Abstract
Objective:
The objective of this study was to provide up-to-date estimates of the clinical and economic burden that occurs during inpatient treatment of cancer patients with febrile neutropenia (FN).
Methods:
A retrospective cohort study was conducted using 2007–2010 hospital discharge data from the Premier database. The study population included adult patients with discharge diagnoses of neutropenia (ICD-9 code 288.0x) with fever or infection and receipt of intravenous antibiotics and female breast cancer, lung cancer, colorectal cancer, ovarian cancer, non-Hodgkin lymphoma (NHL), or Hodgkin lymphoma. Primary study outcomes were inpatient mortality, hospital length of stay (LOS), and total hospitalization cost for each patient’s first FN-related hospitalization. Logistic regressions (for mortality) and multivariate linear regressions (for LOS and cost) were conducted to assess the effect of comorbidities and infection types on study outcomes, adjusting for other patient and hospital characteristics.
Results:
Among 16,273 cancer patients hospitalized with FN, the inpatient case fatality rate was 10.6%, mean LOS was 8.6 days, and mean total hospitalization cost was $18,880. Lung cancer patients had the highest inpatient case fatality rate (15.7%), and NHL patients had the longest LOS (10.1 days) and the highest cost ($24,218). Multivariate analyses showed that most comorbidities were associated with a greater risk of mortality, longer LOS, and higher cost. Septicemia/bacteremia and pneumonia were associated with a greater risk of mortality, and most types of infection were associated with a longer LOS and higher cost.
Limitations:
The total burden of FN may be under-estimated in this study because outpatient treatment and any patient deaths or costs that occurred outside of Premier hospitals could not be captured.
Conclusions:
FN-related hospitalizations among cancer patients are costly and accompanied by considerable mortality risk. Substantial differences in the clinical and economic burden of FN exist depending on cancer types, comorbidities, and infection types.
Introduction
Chemotherapy-induced febrile neutropenia (FN) is a common, life-threatening side-effect of myelosuppressive chemotherapyCitation1,Citation2 that often requires immediate hospitalization and administration of empiric, broad-spectrum antibioticsCitation3. Each year, conservative estimates project that 60,000–100,000 cancer patients in the US are hospitalized with neutropenic complicationsCitation4.
Significant risk of mortality and substantial costs are often seen during hospitalization of cancer patients with FNCitation3,Citation5,Citation6. The clinical and economic burden of FN-related hospitalizations among cancer patients have been examined in two large US studies. Using discharge data from hospital databases from seven states in 1999, Caggiano et al.Citation5 reported an inpatient case fatality rate of 6.8%, mean hospital length of stay (LOS) of 9.2 days, and mean total cost for hospitalization of $13,372 (1999 US dollars). Similarly, Kuderer et al.Citation3 used hospital discharge data from 1995–2000 from 115 academic medical centers and reported an inpatient case fatality rate of 9.5%, mean hospital LOS of 11.5 days, and mean total cost for hospitalization of $20,290 (2000 US dollars). In a more recent study (2005–2008), Schilling et al.Citation6 used a hospital database maintained by ASPEN Healthcare Metrics and reported an inpatient case fatality rate of 13.7%, a mean LOS of 10.7 days, and mean hospitalization cost of $22,839 (2009 US dollars) for cancer patients with neutropenia and fever or infection. However, the size of the Schilling et al. study was relatively small (n = 1809) compared with the previous studies (n = 20,780 and n = 41,779)Citation3,Citation5,Citation6.
Cost data from these previous studies are now more than 10 years old or based on a relatively small study size. Additionally, clinical management of FN has changed considerably with incorporation of new antimicrobial drugs, better tailoring of antimicrobial therapy to the risk of complications, and increased outpatient management of low-risk FN patientsCitation7–9. These changes may affect the clinical and economic burden of FN-related hospitalizations. This retrospective cohort study used discharge data of cancer patients hospitalized with FN from one of the largest hospital databases in the US to provide up-to-date information on the clinical and economic burden of FN.
Patients and methods
Study population
This retrospective cohort study included adult patients ≥18 years of age with FN and a primary cancer type of female breast cancer, lung cancer, colorectal cancer, ovarian cancer, non-Hodgkin lymphoma (NHL), or Hodgkin lymphoma who were discharged from January 1, 2007–December 31, 2010 from a US hospital participating in the database maintained by Premier. Patients were excluded if they had received a hematopoietic stem cell transplant at any time before or during the index hospitalization or if they had diagnoses of multiple primary cancer types based on relevant Current Procedural Technology (CPT), International Classification of Disease, 9th edition (ICD-9), or Healthcare Common Procedure Coding System (HCPCS) codes.
FN was identified based on a discharge diagnosis of neutropenia (principal or secondary ICD-9 diagnosis code 288.0x), fever (principal or secondary ICD-9 diagnosis code 780.6x), or infection (codes listed in Supplemental Table 1), and receipt of any intravenous antibiotic agent recommended by the Infectious Disease Society of America (IDSA)Citation8,Citation10 (Supplemental Table 2) for initial empirical therapy. Initial empirical therapy was defined as the receipt of such agents on 2 or more consecutive days during the hospitalization (or anytime before death if death occurred within 1 day after admission) when the first injection of such agent(s) occurred within the first 5 days after admission. Cancer type was ascertained based on a corresponding discharge diagnosis ICD-9 code (Supplemental Table 1).
Premier database
The Premier database includes extensively validated discharge files from all inpatients and visit records of hospital-based outpatients from over 400 geographically diverse US hospitals. Compared with the 2007 American Hospital Association (AHA) statisticsCitation11, hospitals covered by Premier’s database in 2008 were more likely to have larger size (300+ beds), be located in the South rather than the Northeast region, and be teaching hospitals. In addition to the data elements available in most standard hospital discharge files (e.g., demographics, diagnoses, discharge status, and physician and hospital characteristics), the Premier database also contains a date-stamped log of all cost items including procedures, medications, laboratory, and diagnostic and therapeutic services at the individual patient level. Data were fully de-identified and compliant with the 1996 Health Insurance Portability and Accountability Act (HIPAA).
Study outcomes
For any cancer patient with multiple FN-related hospitalization episodes during the study period, the first hospitalization episode of the patient (index hospitalization) was selected for the analysis. The primary study outcomes were inpatient mortality, hospital LOS, and total hospitalization cost, all of which were based on the index hospitalization. Mortality risk was reported as a simple inpatient case fatality rate (number of deaths divided by the number of admissions). All LOS calculations were based on the relevant admission and discharge dates. Total hospitalization cost was determined from clinical and billing records. All costs represent the hospital’s internal assessment of the actual cost to the hospital of delivering goods and services (not amount charged or reimbursed) and were reported to Premier in accordance with accepted accounting standards. These costs were not further standardized or adjusted when recorded in the Premier database. However, for the analyses presented here, costs from the database were adjusted to 2010 US dollars according to the hospital and related services component of the Consumer Price Index (CPI). Patient’s discharge and survival outcomes (discharged alive, died before being discharged, or still in hospital and alive) on each day within 30 days after the start of the index hospitalization were also examined. Additionally, patient demographics, patient clinical characteristics, hospital characteristics, and hospitalization characteristics were summarized for each hospital episode included in the study.
Secondary outcomes included use of antimicrobial agents, detailed components of cost and resource use (e.g., use of the intensive care unit [ICU] and ICU LOS), and incidence, cost, and clinical outcomes for FN-related re-hospitalizations. Only re-hospitalizations more than 2 days after discharge from the index hospitalization were examined as re-admission outcomes. Re-admission within 2 days of the discharge from the index hospitalization was considered as an extended part of the index hospitalization.
Statistical analyses
Means, medians, standard deviations (SD), and 95% confidence intervals (CI) were reported as appropriate for continuous variables, and percentages and 95% CI were reported for all indicator variables. Descriptive analyses were used to summarize the mortality risk outcomes, utilization, cost, and all the other study variables (patient demographic and clinical characteristics, hospital characteristics, and hospitalization characteristics). All analyses were undertaken for the overall patient population, by whether the patient died during the index hospitalization, and by cancer type (female breast cancer, lung cancer, colorectal cancer, ovarian cancer, NHL, and Hodgkin lymphoma).
Pooled analysis
Summary statistics for the primary outcomes of the study (inpatient mortality risk, hospital LOS, and total hospitalization cost) were reported for index hospitalizations with different characteristics at the patient, hospital, or hospitalization level.
Multivariate analysis
Multivariate regression analysis was conducted separately for female breast cancer, lung cancer, and NHL to quantify the effect of patient comorbidity (i.e., congestive heart failure, other heart disease, lung disease, liver disease, renal disease, diabetes, cerebrovascular disease, peripheral vascular disease, deep venous thrombosis, pulmonary embolism, and anemia) and infection type (i.e., septicemia/bacteremia, pneumonia, urinary tract infection, intravenous site infection, candidiasis, bacterial infection–site unspecified, and other miscellaneous infection) on the primary study outcomes. Comorbidities were defined on the basis of discharge diagnosis from the index hospitalization and any previous hospitalizations within 180 days prior to the index hospitalization. Infection types were defined on the basis of discharge diagnosis from the index hospitalization. ICD-9 codes used to identify comorbidities and types of infection are listed in Supplemental Table 1. Logistic regression was used to estimate inpatient mortality risk, and ordinary least squares linear regression was used to estimate hospital LOS and total hospitalization cost. Each model included two alternative specifications. The first included indicator variables for comorbidities of interest; the second included the total number of comorbidities. The following potentially confounding variables were controlled: patient characteristics (i.e., age, gender, race/ethnicity, and primary payer) and hospitalization characteristics (i.e., discharge year and admission source). Hospital characteristics (i.e., region, urban/rural, teaching status, and bed size) were also controlled to account for variability in costs of FN due to region and type of hospital.
Linked claims data analysis
To estimate the percentage of hospitalizations for FN that were preceded by chemotherapy use in the 30 days before the FN episode, the Premier database was linked to the OptumInsight database, a large outpatient research database that incorporates de-identified medical and pharmacy claims, lab results, and enrollment data covering more than 35 million patients for a national managed care population. The linking process required hospital-level matching (based on the hospital’s Medicare provider number and other hospital details) and discharge-level matching (based on admission date, discharge data, DRG [diagnosis related groups] or MS-DRG [Medicare severity diagnosis related groups], patient gender, and patient birth date). Only discharges with exactly matched records in both databases could be used in the linked claims data analysis. Previous analysis has shown that 2.7% of discharges with cancer as the principal ICD-9 diagnosis in the Premier database were linked to the OptumInsight databaseCitation12, which is likely a result of different populations captured in each database. The OptumInsight database includes a single payer and represents commercially-insured individuals, who tend to be younger. Older patients, in whom cancer is more prevalent, are less likely to be covered in any commercial insurance database. The Premier database represents all payers and is more likely to capture the older patients.
Results
Patient demographics and hospitalization characteristics
A total of 16,273 index hospitalizations for adult cancer patients with evidence of neutropenia and fever/infection and administration of intravenous antibiotics were identified in the Premier database (Supplemental Table 3). Overall, patients had a mean (SD) age of 62.7 (13.5) years, and 60.1% of patients were female. The most common primary cancer types identified were NHL (n = 5437; 33.4%), lung cancer (n = 4792; 29.4%), and female breast cancer (n = 3279; 20.1%). Most patients had two or more comorbidities (n = 10,384; 63.8%), and the most common comorbidities were anemia (n = 10,102; 62.1%), lung disease (n = 6037; 37.1%), heart disease (congestive heart failure: n = 1217; 7.5% and other heart disease: n = 9441; 58.0%), and renal disease (n = 3392; 20.8%). Additional patient demographics are shown in .
Table 1. Patient sample.
Approximately half of the patients were treated by an attending physician with an oncology specialty (n = 7937; 48.8%). All patients had some type of infection, with septicemia/bacteremia (n = 4657; 28.6%) and pneumonia (n = 3552; 21.8%) being the most common types specified. Consistent with the study definition of FN, all patients received antibiotics. A total of 6666 patients (41.0%) received antifungals and 2822 (17.3%) received antivirals. Additional hospitalization characteristics are shown in .
Hospital providers were geographically distributed across the US, with 2955 patients (18.2%) treated at hospitals in the Northeast, 2709 (16.6%) treated at hospitals in the West, 7131 (43.8%) treated at hospitals in the South, and 3478 (21.4%) treated at hospitals in the Midwest. The majority of patients were treated at hospitals in urban locations (n = 14,558; 89.5%), with only a small sub-set treated at rural hospitals (n = 1715; 10.5%). Nearly half of the patients were treated at a teaching hospital (n = 7263; 44.6%). Most patients were treated at larger hospitals, with 36.6% of patients treated at hospitals having 300–499 beds, and 36.6% of patients treated at hospitals with 500+ beds.
Clinical and economic outcomes
Overall, 14,555 patients (89.4%) were discharged alive. Most patients were discharged to home (n = 12,273; 75.4%). The remainder of patients were discharged to another healthcare facility (n = 2140; 13.2%) or discharged to a different or unknown destination (n = 142; 0.9%).
Altogether 1718 patients died; the inpatient case fatality rate was 10.6% (95% CI: 10.1–11.0) overall and differed among cancer types (). The inpatient case fatality rate was highest for patients with lung cancer (n = 750; 15.7%; 95% CI = 14.6–16.7) and lowest for patients with female breast cancer (n = 182; 5.6%; 95% CI = 4.8–6.3). At the end of 30 days after admission to the hospital, 86.8% of patients had been discharged alive, 3.5% were still hospitalized, and 9.8% had died before being discharged ().
Figure 1. Mortality and discharge outcomes by day since admission. The percentages on the right margin of each graph represent the proportion of patients classified as “discharged alive,” “still in hospital and alive,” and “died before being discharged” during the 30 days following admission. For patients who were discharged alive, survival status after discharge date is unavailable in the hospital database.
Table 2. Economic and clinical outcomes.
For the index hospitalization, mean LOS across all cancer types was 8.6 days (95% CI = 8.5–8.8). A total of 3101 patients (19.1%) were treated in an ICU setting during their index hospitalization, with a mean LOS of 5.2 days spent in ICU. Hospital LOS varied among cancer types (). Patients with NHL had the longest mean LOS (10.1 days; 95% CI = 9.8–10.4), and patients with female breast cancer had the shortest mean LOS (5.9 days; 95% CI = 5.7–6.1).
Total hospitalization cost for the index hospitalization was available for 16,268 patients. Mean hospitalization cost across all cancer types was $18,880 (95% CI = 18,479–19,281); the mean cost per day of hospitalization was $2169 (95% CI = 2150–2189). Consistent with hospital LOS, cost was variable based on cancer type. NHL had the highest mean cost ($24,218; 95% CI = 23,328–25,109), and female breast cancer had the lowest mean cost ($11,132; 95% CI = 10,649–11,615). However, mean cost per day was similar among cancer types ($1901–$2348). Detailed components of hospital costs are available in .
Mean total hospitalization cost was lower and LOS was shorter for patients who were discharged alive than for patients who were discharged dead (). For patients discharged alive, mean cost was $17,322 (95% CI = 16,939–17,704) and mean LOS was 8.3 days (95% CI = 8.2–8.5). For patients who died while they were in the hospital, mean cost was $32,088 (95% CI = 30,219–33,956), and mean LOS was 11.0 days (95% CI = 10.4–11.6).
Table 3. Cost and LOS by discharge status.
Re-admissions
Re-admission to the hospital was fairly common. In the 30 days following hospital discharge, 3460 patients (23.8%) were re-admitted to the hospital for any reason, and 853 patients (5.9%) were re-admitted to the hospital for FN-related reasons (). The FN-related re-admission rate was higher for patients with NHL (9.9%) and for patients with Hodgkin lymphoma (8.6%) than for patients with other tumor types (2.3–4.1%).
A total of 2220 patients (15.3%) were re-admitted for FN-related reasons at any time. For re-admissions among these patients, the inpatient case fatality rate was 7.5% (n = 167; 95% CI = 6.4–8.6), mean LOS was 8.0 days (95% CI = 6.4–8.6), and mean cost was $17,235 (95% CI = 16,128–18,342).
Pooled analysis
When data were pooled across all cancer types studied and analyzed based on different sub-groups (e.g., by patient age), several factors were associated with increased mortality, LOS, and/or cost (). Of note, older age was associated with a higher mortality, with patients 18–44 years of age having an inpatient case fatality rate of 5.1% (95% CI = 4.0–6.1) and patients ≥75 years of age having an inpatient case fatality rate of 15.8% (95% CI = 14.6–17.1). However, LOS and cost were comparable among age groups. Males had higher mortality (12.6% vs 9.2%), a slightly longer LOS (9.1 vs 8.3 days), and higher cost ($21,038 vs $17,447) than did females. Compared with the average across all patients, most specific comorbidities were associated with higher mortality, longer LOS, and higher cost (). Finally, the inpatient case fatality rate in patients with septicemia/bacteremia (25.1%; 95% CI = 23.9–26.4) or pneumonia (20.3%; 95% CI = 19.0–21.6) was higher than the rate across the entire population ().
Table 4. Pooled analysis.
Multivariate analysis
Multivariate analyses were performed for female breast cancer, lung cancer, and NHL because these cancer types had a sufficient number of patients to perform meaningful analyses. First, specific types of comorbidities and infection were evaluated as potential risk factors for inpatient mortality and higher economic burden (i.e., longer LOS and higher hospitalization cost). Most comorbidities were associated with a higher risk of in-hospital mortality (), longer LOS (), and higher cost (). For example, for patients with NHL and lung disease, the risk of mortality was higher (risk ratio [RR] as approximated by the odds ratio = 4.5; 95% CI = 3.6–5.7), LOS was 3.6 days longer (95% CI = 3.0–4.2), and cost was $13,268 higher (95% CI = 11,441–15,095) than in NHL patients without lung disease. Similarly, for patients with NHL and liver disease, the risk of mortality was higher (RR = 2.3; 95% CI = 1.6–3.2), LOS was 4.7 days longer (95% CI = 3.6–5.9), and cost was $14,634 higher (95% CI = 11,239–18,029) than in NHL patients without liver disease. For patients with NHL and renal disease, the risk of mortality was higher (RR = 3.1; 95% CI = 2.5–3.8), LOS was 2.3 days longer (95% CI = 1.7–3.0), and cost was $10,408 higher (95% CI = 8391–12,425) than in NHL patients without renal disease. Similar results were seen for patients with other cancer types ().
Table 5. Multivariate analysis of in-hospital mortality by specific comorbidities and infection types.
Table 6. Multivariate analysis of length of stay by specific comorbidities and infection types.
Table 7. Multivariate analysis of hospitalization cost by specific comorbidities and infection types.
Septicemia/bacteremia and pneumonia were also associated with higher risk of mortality (), longer LOS (), and higher cost () across all three cancer types studied. For patients with female breast cancer who had septicemia or bacteremia, the risk of mortality was higher (RR = 4.1; 95% CI = 2.6–6.5), LOS was 1.7 days longer (95% CI = 1.0–2.3), and cost was $5664 higher (4233–7095) than for patients with female breast cancer who did not have septicemia or bacteremia. Similarly, for female breast cancer patients with pneumonia, the risk of mortality was higher (RR = 2.1; 95% CI = 1.3–3.3), LOS was 2.5 days longer (95% CI = 1.8–3.2), and cost was $6593 higher (95% CI = 4949–8237) than for patients with female breast cancer who did not have pneumonia.
When the number of comorbidities was included in the multivariate models rather than specific comorbidities, higher numbers of comorbidities were associated with higher risk of mortality and higher cost. For example, for patients with female breast cancer, the risk of mortality was greater for patients with two comorbidities (RR = 3.5; 95% CI = 1.5–8.1) than for patients with no comorbidities. The risk of mortality continued to increase as the number of comorbidities increased. Relative to patients with no comorbidities, the RR of mortality for female breast cancer patients with three comorbidities was 5.2 (95% CI = 2.2–12.2), and the RR for patients with four or more comorbidities was 9.6 (95% CI = 4.0–22.7). Mean cost for patients with NHL who had one comorbidity was $4084 higher (95% CI = 1107–7061) than cost for NHL patients who had no comorbidities. Similarly, cost for patients with NHL who had two comorbidities was $9627 higher (95% CI = 6626–12,628), cost for patients with three comorbidities was $16,949 higher (95% CI = 13,765–20,133), and cost for patients with four or more comorbidities was $28,768 higher (95% CI = 25,429–32,107) than cost for patients with no comorbidities.
Linked claims data analysis
A total of 371 records (2.3% of all discharges in the study sample) from the Premier database could be linked to the OptumInsight database, with 105 patients with female breast cancer, 86 patients with lung cancer, 41 patients with colorectal cancer, 14 patients with ovarian cancer, 113 patients with NHL, and 12 patients with Hodgkin lymphoma having records in both databases. Most patients were documented to have received chemotherapy within 30 days before the index hospitalization for FN (n = 291; 78.4%). The percentage of patients who were documented to have received chemotherapy in the 30 days before their index hospitalization for FN was highest for patients with female breast cancer (90.5%; n = 95) and lowest for patients with NHL (59.3%; n = 67). Similar percentages of patients had prior chemotherapy among patients with lung cancer (82.6%; n = 71), colorectal cancer (87.8%; n = 36), ovarian cancer (85.7%; n = 12), and Hodgkin lymphoma (83.3%; n = 10).
Discussion
In this study of 16,273 cancer patients hospitalized with FN, the average inpatient case fatality rate for patients with all cancer types we studied was 10.6%, LOS was 8.6 days, and cost of hospitalization was $18,880. Several factors were associated with variability in these measures, including cancer type, discharge status, presence of comorbidities, and type of infection. Of note, LOS was longer and cost was higher among patients who died while hospitalized than among patients discharged alive. These results are consistent with Michels et al.Citation13, who reported that, among FN patients, those who died had higher mean per patient per month total cost than surviving FN patients ($21,214; 95% CI = 19,192–23,237 vs $8227; 95% CI = 7987–8466).
This study provides updated estimates of the inpatient case fatality rates, LOS, and cost that accompany FN treated in the hospital setting. Two large studies of US cancer patients conducted a decade ago reported inpatient case fatality rates of 6.8% and 9.5%, mean LOS of 9.2 days and 11.5 days, and mean total cost of hospitalization of $13,400 (1999 US dollars) and $20,290 (2000 US dollars)Citation3,Citation5. In a more recent study (2005–2008), the inpatient case fatality rate was 13.7%, mean LOS was 10.7 days, and mean hospitalization cost was $22,839 (2009 US dollars)Citation6. Differences in the cancer types included in each study population may have contributed to differences seen among the studies. For example, treatment for patients with hematological cancers was generally accompanied by higher cost and a longer LOS than for patients with solid tumors, and the inpatient case fatality rate is often much greater among patients with lung cancer than among patients with female breast cancerCitation3,Citation5,Citation6. The definitions of FN, healthcare facility types, patient comorbidities, types of infection, and changes in cost of care and treatment of FN over time may also have contributed to the differences seen among studies.
In addition to providing updated estimates on the impact of FN, several other factors differentiate this study. The National Comprehensive Cancer Network (NCCN)Citation14, European Organisation for Research and Treatment of Cancer (EORTC)Citation15, and Infectious Disease Society of America (IDSA) guidelinesCitation8 all recommend prompt treatment of FN with broad-spectrum antibiotics. In light of these recommendations, receipt of intravenous antibiotics was incorporated into the definition of FN for this study, leading to a more refined definition of FN. Additionally, this study provides considerable detail on the economic and clinical burden of FN, including detailed cost components and resource utilization measures, day-by-day patient survival, and the incidence, cost, and the inpatient case fatality rate during re-admission. Finally, mean hospitalization cost in this study was determined based on the actual costs reported by each hospital rather than costs derived from charges (under certain assumption of cost-to-charge ratio), which were used in earlier studiesCitation3,Citation5. Together, these details provide a more comprehensive assessment of the clinical and economic impact of FN than in previous studies.
This study also evaluated the impact of comorbidities and type of infections on mortality, LOS, and cost. The results from the pooled analysis and the multivariate analyses were similar for most comorbidities. However, in the pooled analysis, the inpatient case fatality rate for anemia (11.0%) was higher than the inpatient case fatality rate across all patients (10.6%), while the multivariate analyses indicated that anemia might be associated with lower risk of mortality (see ). Several factors may have contributed to this discrepancy. First, the pooled analysis looked at results across all major cancer types examined in this study, while the multivariate analyses were conducted separately for female breast cancer, lung cancer, and NHL. Additionally, other comorbidities, infections, or other variables could confound the relationship between anemia and mortality.
The patients in this study may represent a population that is at high risk for serious complications of FN. Patients with FN can be categorized as high- or low-risk on the basis of validated risk modelsCitation16,Citation17. Low-risk patients are candidates for oral antibiotics in the inpatient setting or outpatient management of FNCitation14,Citation16–20 and would not be captured in this study population. Little information is available about the incidence and treatment of low-risk patients, but recent estimates suggest ∼20% of patients may be treated for FN in the outpatient settingCitation21–23.
Hospitalization with intravenous antibiotics is the current standard of care for FN, but the clinical and economic burden of FN extends beyond the initial hospitalization. Among patients with FN, subsequent neutropenia-related care has been estimated to represent ∼40% of the total healthcare costs for treating FNCitation2. Finally, indirect costs of FN, such as lost productivity, care-giving burden, and cost of transportation to and from the healthcare facility, can increase cost estimates of FNCitation24–26. These costs and changing treatment patterns should be considered when determining the impact of FN.
Determining the true cost of FN is an important factor in clinical decision-making, and estimates of FN cost can impact patient care. For example, initial estimates of the hospitalization cost for FN were $1000 per dayCitation27. In this setting, colony-stimulating factor (CSF) use was predicted to be cost-saving when the risk of hospitalization with FN was >40%Citation27. More recent estimates that include a broader range of costs predicted that CSFs would be cost-saving when the risk of hospitalization with FN was ∼20%Citation1,Citation28. These estimates are consistent with current NCCN and ASCO guidelines for use of CSFs to reduce the risk, duration, and severity of FNCitation29,Citation30. These guidelines recommend prophylactic use of CSFs in patients with a ≥20% risk of FN based on the chemotherapy regimen and treatment-related factors. Careful consideration of the risk and costs of FN is important to help inform appropriate and cost-effective patient care.
This study used inpatient data from over 400 hospitals included in the database maintained by Premier. A large number of cancer patients hospitalized with FN were identified, and data were extensively validated. One key limitation of this study is the possible under-estimation of the burden of FN because no outpatient management of FN was captured, any costs or patient deaths that occurred outside of Premier hospitals were not captured, and only re-admissions to the same facility as the index hospitalization could be identified in Premier’s database. Additionally, absolute neutrophil count (ANC) and oral body temperature were not available in Premier’s database, and the clinical definition of FN could not be used. Furthermore, no single ICD-9 code exists for FN, which can contribute to errors of omission and commission during coding of the data. As an operational definition of FN, hospitalization with a diagnosis of neutropenia has a sensitivity of 67–80% and a specificity of 94%Citation31,Citation32 when compared with the clinical definition of FN, which is fever (a single oral temperature ≥38.3°C or ≥38.0°C for at least 1 hour) with neutropenia (<500 neutrophils/µL or <1000 neutrophils/µL and a predicted decline to <500 neutrophils/µL over the next 48 hours)Citation14. To further validate the definition of FN, Premier records were linked to the OptumInsight database to determine the percentage of patients that had received chemotherapy before hospitalization for FN. Only 371 Premier records could be linked to the OptumInsight database. Additionally, the OptumInsight data extract used in the analysis might not comprehensively capture oral chemotherapy drugs, which could lead to under-representation of the percentage of patients who received chemotherapy, especially for patients with NHL. However, for all tumor types examined, except for NHL, 82.6–90.5% patients had evidence of chemotherapy within 30 days before the index hospitalization, which provides additional support for the validity of our FN definition.
Conclusion
FN-related hospitalizations among cancer patients remain costly and are accompanied by considerable mortality risk. Substantial differences in the clinical and economic burden of FN exist depending on type of cancer, comorbidities, and type of infection.
Transparency
Declaration of funding
This study was sponsored by Amgen Inc.
Declaration of financial/other relationships
X. Li, R. L. Barron, and J. C. Legg are employees of and stockholders in Amgen Inc. J. A. Gayle and F. R. Ernst are employees of Premier healthcare alliance, which received funding from Amgen Inc. B. Dulisse was an employee of Premier healthcare alliance at the time this study was conducted. K. J. Rothman and J. A. Kaye are employees of RTI Health Solutions, an independent, non-profit research organization which was engaged by Amgen Inc. to consult on the design of the study and interpretation of the results. JME Peer Reviewers on this manuscript have no relevant financial relationships to disclose.
Supplementary Material
Download PDF (48.1 KB)Acknowledgments
The authors thank Dr. Gary H. Lyman for kindly providing the ICD-9 codes for comorbidities and infection types used by Kuderer et alCitation3. Kerri Hebard-Massey (Amgen Inc.) provided medical writing support.
References
- Lyman GH, Kuderer N, Greene J, et al. The economics of febrile neutropenia: implications for the use of colony-stimulating factors. Eur J Cancer 1998;34:1857-64
- Weycker D, Malin J, Edelsberg J, et al. Cost of neutropenic complications of chemotherapy. Ann Oncol 2008;19:454-60
- Kuderer NM, Dale DC, Crawford J, et al. Mortality, morbidity, and cost associated with febrile neutropenia in adult cancer patients. Cancer 2006;106:2258-66
- Kozma CM, Dickson M, Chia V, et al. Trends in neutropenia-related inpatient events. J Oncol Pract 2012;8:149-55
- Caggiano V, Weiss RV, Rickert TS, et al. Incidence, cost, and mortality of neutropenia hospitalization associated with chemotherapy. Cancer 2005;103:1916-24
- Schilling MB, Parks C, Deeter RG. Costs and outcomes associated with hospitalized cancer patients with neutropenic complications: a retrospective study. Exp Ther Med 2011;2:859-66
- de Naurois J, Novitzky-Basso I, Gill MJ, et al. Management of febrile neutropenia: ESMO Clinical Practice Guidelines. Ann Oncol 2010;21(5 Suppl):v252-6
- Freifeld AG, Bow EJ, Sepkowitz KA, et al. Clinical practice guideline for the use of antimicrobial agents in neutropenic patients with cancer: 2010 update by the infectious diseases society of america. Clin Infect Dis 2011;52:e56-93
- Klastersky J, Awada A, Paesmans M, et al. Febrile neutropenia: a critical review of the initial management. Crit Rev Oncol Hematol 2011;78:185-94
- Hughes WT, Armstrong D, Bodey GP, et al. 2002 guidelines for the use of antimicrobial agents in neutropenic patients with cancer. Clin Infect Dis 2002;34:730-51
- American Hospital Association. 2007 AHA Hospital Statistics. Chicago, IL: Health Forum, LLC. 2006.
- Ernst FR, Davis TE, Ryan AC, et al. Linkage of a multi-hospital database to health plan claims data for continuum of care outcomes research [Abstract and Poster]. Orlando, FL: International Society for Pharmacoeconomics and Outcomes Research (ISPOR), 2009
- Michels SL, Barron RL, Reynolds MW, et al. Costs associated with febrile neutropenia in the US. Pharmacoeconomics 2012;30:809-23
- Prevention and treatment of cancer-related infections, v1.2012. NCCN Clinical Practice Guidelines in Oncology, 2012. Available at: http://www.nccn.org/professionals/physician_gls/pdf/infections.pdf [Last accessed 20 December 2012]
- Aapro MS, Bohlius J, Cameron DA, et al. 2010 update of EORTC guidelines for the use of granulocyte-colony stimulating factor to reduce the incidence of chemotherapy-induced febrile neutropenia in adult patients with lymphoproliferative disorders and solid tumours. Eur J Cancer 2011;47:8-32
- Klastersky J, Paesmans M, Rubenstein EB, et al. The Multinational Association for Supportive Care in Cancer risk index: a multinational scoring system for identifying low-risk febrile neutropenic cancer patients. J Clin Oncol 2000;18:3038-51
- Talcott JA, Finberg R, Mayer RJ, et al. The medical course of cancer patients with fever and neutropenia. Clinical identification of a low-risk subgroup at presentation. Arch Intern Med 1988;148:2561-8
- Klastersky J, Paesmans M, Georgala A, et al. Outpatient oral antibiotics for febrile neutropenic cancer patients using a score predictive for complications. J Clin Oncol 2006;24:4129-34
- Talcott JA, Siegel RD, Finberg R, et al. Risk assessment in cancer patients with fever and neutropenia: a prospective, two-center validation of a prediction rule. J Clin Oncol 1992;10:316-22
- Talcott JA, Yeap BY, Clark JA, et al. Safety of early discharge for low-risk patients with febrile neutropenia: a multicenter randomized controlled trial. J Clin Oncol 2011;29:3977-83
- Stokes ME, Muehlenbein CE, Marciniak MD, et al. Neutropenia-related costs in patients treated with first-line chemotherapy for advanced non-small cell lung cancer. J Manag Care Pharm 2009;15:669-82
- Weycker D, Edelsberg J, Kartashov A, et al. Risk and healthcare costs of chemotherapy-induced neutropenic complications in women with metastatic breast cancer. Chemotherapy 2012;58:8-18
- Weycker D, Barron R, Kartashov A, et al. Incidence, treatment, and consequences of chemotherapy-induced febrile neutropenia in inpatient and outpatient settings [Abstract and Poster]. Atlanta, GA: American Society of Hematology Annual Meeting, 2012
- Bennett CL, Calhoun EA. Evaluating the total costs of chemotherapy-induced febrile neutropenia: results from a pilot study with community oncology cancer patients. Oncologist 2007;12:478-83
- Cosler LE, Calhoun EA, Agboola O, et al. Effects of indirect and additional direct costs on the risk threshold for prophylaxis with colony-stimulating factors in patients at risk for severe neutropenia from cancer chemotherapy. Pharmacotherapy 2004;24:488-94
- Hendricks AM, Loggers ET, Talcott JA. Costs of home versus inpatient treatment for fever and neutropenia: analysis of a multicenter randomized trial. J Clin Oncol 2011;29:3984-9
- Lyman GH, Lyman CG, Sanderson RA, et al. Decision analysis of hematopoietic growth factor use in patients receiving cancer chemotherapy. J Natl Cancer Inst 1993;85:488-93
- Cosler LE, Sivasubramaniam V, Agboola O, et al. Effect of outpatient treatment of febrile neutropenia on the risk threshold for the use of CSF in patients with cancer treated with chemotherapy. Value Health 2005;8:47-52
- Myeloid growth factors, v1.2012. NCCN Clinical Practice Guidelines in Oncology, 2012. Available at: http://www.nccn.org/professionals/physician_gls/pdf/myeloid_growth.pdf [Last accessed 20 December 2012]
- 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;24:3187-205
- Chen-Hardee S, Chrischilles EA, Voelker MD, et al. Population-based assessment of hospitalizations for neutropenia from chemotherapy in older adults with non-Hodgkin's lymphoma (United States). Cancer Causes Control 2006;17:647-54
- Weycker D, Sofrygin O, Seefeld K, et al. Technical evaluation of methods for identifying chemotherapy-induced febrile neutropenia in healthcare claims databases. BMC Health Serv Res 2013;13:60