Economic burden in US patients with newly diagnosed acute myeloid leukemia receiving intensive induction chemotherapy

Acute myeloid leukemia (AML) is a rare malignancy with a poor prognosis [1]. In the United States, in 2020, the incidence rate of AML was estimated at 4.3 per 100,000 individuals, and the death rate was approximately 2.8 per 100,000 individuals [1]. AML is most frequently diagnosed in individuals aged 65–74 years, with a median age at diagnosis of 68 years and the median age at death of 73 years [1]. While sometimes curable, overall outcomes in AML are poor, and most older patients with AML will die from this disease [2].

The current standard of care for AML consists of induction chemotherapy to reduce the leukemic burden and produce complete remission, followed by consolidation therapy to eradicate residual disease and maintain remission. Standard induction therapy consists of cytarabine and anthracycline, while post-remission therapy comprises consolidation chemotherapy alone, hematopoietic stem cell transplantation (HSCT) alone, or consolidation chemotherapy followed by HSCT [3]. Recent clinical trials suggest that maintenance therapy with injectable or oral azacitidine for patients with AML who have achieved complete remission after intensive induction chemotherapy may extend disease-free survival compared with no further treatment and best supportive care [4–8]. Oral azacitidine is the only maintenance treatment to show an overall survival benefit in this population [9]. Despite improvements in available treatment options, most patients (50–80%) eventually relapse following induction and/or post-remission therapies [9,10] and 5-year survival post-relapse is low, at approximately 10% [11].

Retrospective claims analyses indicate that AML is associated with substantial healthcare resource utilization (HCRU) and costs in the USA [12–16], where costs of each treatment episode across the course of the disease were estimated to range from US$145,189 to $198,657 for induction therapy, $28,137 to $73,428 for consolidation, with the cost of HSCT ranging from $219,973 to $329,621 [15,16]. Inpatient hospitalization costs are the primary cost drivers in newly diagnosed AML, as patients remain hospitalized until blood cell count recovery following completion of chemotherapy, which requires continuous patient monitoring and rapid treatment of adverse events [15,17,18]. Outpatient and emergency department (ED) visits are also common; an estimated 70%–75% of patients with newly diagnosed AML may have ≥1 unplanned ED visit or hospitalization and nearly all may have outpatient days [14,18]. High-intensity therapies for AML carry a risk of complications requiring intensive care unit (ICU)-level care, and a quarter of all patients with AML are admitted to the ICU [19].

Characterizing hospital-based HCRU and costs in patients with newly diagnosed AML is important for understanding the burden of AML subsequent to diagnosis and the potential healthcare system benefits of existing and future therapies for AML. Previous studies have used adjudicated claims data to estimate the economic burden among patients with newly diagnosed AML in the US [12,14,17]. Adjudicated claims data correspond to payment for health services delivered by healthcare providers and facilities. Adjudicated claims data may underestimate HCRU and costs, as they reflect reimbursement and only represent the amount an insurance company pays. Adjudicated claims data are dependent on patient cost-sharing requirements, such as deductibles, co-payments, and co-insurance amounts, and they lack information on claim rejections or denials of healthcare costs [20]. Electronic healthcare databases fill evidence gaps left by adjudicated claims data, as they offer more accurate and complete information on the HCRU and costs incurred in real-world clinical practice [21]. Electronic healthcare databases contain patient-centered records that include medical and treatment histories and clinical data [22]. Electronic healthcare databases provide extensive, unbiased data reflecting medical care in the general population, representing robust evidence- and population-based research tools [21].

The objective of this study was to utilize clinical and billing data from a large hospital administrative database to comprehensively assess HCRU and costs for patients with newly diagnosed AML receiving intensive induction chemotherapy in the USA.

Materials & methods

Data source

This study used data from the PINC AI™ Healthcare Database (PHD) (formerly known as Premier Healthcare Database) a hospital administrative database, to identify patients with newly diagnosed AML. The PHD is a large, US hospital-based, service-level, geographically diverse, all-payer database that contains administrative, HCRU, and cost data from patient encounters at 1164 contributing hospitals/healthcare systems. Inpatient admissions include over 135 million visits, representing approximately 25% of annual inpatient admissions in the USA. Hospital characteristics, including bed capacity, urban and rural populations served, and teaching status, are recorded in the PHD for the four geographic regions and their respective divisions defined by the US census. The PHD charge master can be used to determine costs (fixed and variable) to a hospital or to a patient's health insurance provider, including hospital services, medical procedures, equipment fees, supplies, drugs, and diagnostic evaluations, as well as days of services delivered and billed [23]. The PHD is an ideal source for studies in patients with newly diagnosed AML, as they undergo hospitalization during induction chemotherapy and may have unplanned ED visits and hospitalizations within the first year of treatment [3].

Study design

This was a retrospective, observational study that assessed HCRU and costs among patients with newly diagnosed AML who received intensive induction chemotherapy. The index date was the date of the first hospitalization/visit with a diagnosis of AML. The pre-index period was the 2 years prior to the index date. The follow-up period from the index date was of variable length, as patients were followed until death, relapse, or last known follow-up, whichever was first. The study period was from 1 July 2014 to 30 September 2019 (Figure 1). A subgroup analysis was conducted in patients who received HSCT on or after the end of the first intensive chemotherapy treatment.

Patient population

The study included patients aged ≥18 years, with an inpatient hospitalization or hospital-based outpatient visit between 1 January 2016, and 31 March 2019, with an AML diagnosis (International Classification of Diseases, tenth edition, clinical modification [ICD-10-CM] diagnosis codes: C92.00, C92.40, C92.50, C92.60, C92.A0), who received CPX-351 or 7 + 3 treatments within 30 days of the index admission, and had ≤280 days from the end of the first induction treatment to remission. This time frame was selected as it allowed inclusion of 93.3% of identified patients. Patients with an AML diagnosis or with a HSCT during the pre-index period were excluded.

Study measures

Patient demographic and clinical characteristics at the index admission, hospital characteristics at the index admission, and hospital-based HCRU and costs between the index date through the end of the follow-up period were assessed (Supplementary Table 1). Chemotherapy-related adverse events were identified by specific ICD-10 codes during the follow-up period.

Patient demographic and clinical characteristics at the index admission included age (groups: 18–44, 45–64, ≥65), sex, race (White, Black, other), primary healthcare coverage payer type (Medicare, Medicaid, commercial, other), Charlson Comorbidity Index (CCI), and individual comorbid conditions in CCI. Time from the end of first induction chemotherapy to remission, and from remission to relapse were recorded.

Hospital characteristics at the index admission included urbanicity, teaching status, geographic region, and bed capacity (grouped as ≤99, 100–199, 200–299, 300–499 and ≥500). Admission type (elective, emergency, trauma center, urgent, other/unknown) and discharge disposition (death, home, hospice, transferred to another healthcare facility, other/unknown) were recorded for each patient visit.

HCRU and costs included total number of outpatient service days per patient, number of inpatient hospitalizations per patient (identified by the principal and/or secondary diagnosis for the admission), inpatient length of stay (LOS) per hospitalization, total inpatient LOS per patient, number of ICU admissions per patient, LOS per ICU admission, total ICU LOS per patient, outpatient hospital costs per visit, total outpatient hospital costs per patient, inpatient hospital costs per hospitalization, total inpatient hospital costs per patient, ICU costs per admission, total ICU costs per patient, and total hospital costs.

Outcomes for patients with HSCT included time from end of first chemotherapy treatment cycle to HSCT, from remission to HSCT, and hospital costs for HSCT visit.

Data analyses

Patient demographics, baseline clinical characteristics, inpatient hospitalizations, ICU admissions, outpatient days, and costs were summarized using unadjusted descriptive analyses. Mean and standard deviation (SD) or median and interquartile range (25th-75th percentiles) for continuous variables, and summed counts and percentages for categorical variables were calculated.

Results

Patient attrition

The study included 642 patients with newly diagnosed AML who received an intensive induction regimen (e.g., liposomal daunorubicin and cytarabine/CPX-351 or 7 + 3 equivalents) (Figure 2).

Patient demographics & characteristics

Mean (median) patient age was 53.7 (56.0) years, 53.1% (n = 341) of patients were male, and 78.2% (n = 502) identified as White. Most patients, 47.2% (n = 303) had commercial insurance, 26.8% (n = 172) had Medicare, 17.6% (n = 113) had Medicaid, and 8.4% (n = 54) had other primary healthcare coverage. Patients had a mean (SD) CCI of 3.5 (2.2). The median (25th-75th percentiles) time from the end of first chemotherapy treatment cycle to remission was 58 (40–90) days, and the median (25th-75th percentiles) time from remission to relapse was 274 (141–389) days (Table 1).

Hospital & admission characteristics

Patients were treated in 129 hospitals across the nation. Among all patients, 95.8% (n = 615) were treated in urban hospitals and 72.6% (n = 466) were treated in teaching hospitals. Most patients were treated in hospitals located in the Atlantic (38.5%, n = 247) and Central 42.4% (n = 272) regions, and 64.3% were treated in hospitals with ≥500 bed capacity (n = 413). The index hospital admission type was emergency in 257 patients (40%) and urgent in 258 patients (40.2%), and 597 patients were discharged home (93%) at index (Table 2).

Adverse events

The most frequently reported adverse events were: 47.2% pyrexia (n = 303), 39.9% neutropenia (n = 256), 28% sepsis (n = 180), 24.3% pneumonia (n = 156), and 17.9 % fungal infection (n = 115) (Table 3).

HCRU

A total of 385 (60%) patients had outpatient service days at a Premier facility, and the median (25th-75th percentiles) number of outpatient service days per patient was 3 (1–6). All patients (100%) had an inpatient hospitalization, and the median (25th-75th percentiles) number of inpatient hospitalizations per patient was 2 (2–3), with a median (25th-75th percentiles) LOS per inpatient hospitalization of 16 (13–21) days. A total of 144 (22.4%) patients had an ICU admission, and the median (25th-75th percentiles) number of ICU admissions per patient was 1 (1–1), with a median (25th-75th percentiles) LOS per ICU admission of 3 (2–8) days, and a total ICU LOS of 3 (2–9) days (Figure 3 and Table 4).

Costs

The median (25th-75th percentiles) cost per outpatient hospital visit was $1,083 ($481–$2,189), and the median (25th-75th percentiles) total outpatient hospital cost was $2,904 ($1,054–$7,217) per patient. The median (25th-75th percentiles) cost per inpatient hospitalization was $34,558 ($25,419–$49,460), and median (25th-75th percentiles) total inpatient hospital cost was $83,440 ($63,067–$113,985) per patient. The median (25th-75th percentiles) cost per ICU visit was $15,771 ($7,209–$27,564), and the median (25th-75th percentiles) ICU cost was $16,550 ($7,368–$36,968) per patient (Table 5 and Figure 4).

HCRU & costs for patients with HSCT

Overall, 92 patients received HSCT. Median (25th-75th percentiles) time from end of first chemotherapy treatment cycle to HSCT was 112 (88–164) days. Median (25th-75th percentiles) time from remission to HSCT was 54 (1–110) days. Median (25th-75th percentiles) total hospital cost for an HSCT visit was $113,638 ($88,547–$183,480) per patient (Table 6).

Discussion

This retrospective analysis used hospital-level data available in the PHD to provide a comprehensive understanding of the hospital-based HCRU and costs in patients with newly diagnosed AML who received intensive induction chemotherapy in the USA. All patients required inpatient hospitalization, and 60% of patients had outpatient service days. Median total hospital cost was $86,584 per patient. The driving cost component was inpatient hospitalization, including median inpatient costs of $34,558 per hospitalization and median total inpatient costs of $83,440 per patient. Median ICU costs were $15,771 per ICU visit, and median total ICU costs were $16,550 per patient. HCRU and costs were highest among patients with newly diagnosed AML who received subsequent HSCT.

The PHD is a comprehensive electronic hospital administrative database that reflects the state of hospital clinical practice in the general US population [23]. The database permitted the creation of a group of patients with newly diagnosed AML who received intensive induction chemotherapy. Evidence derived from real-world data sources such as electronic health records and administrative claims data is increasingly used to improve clinical care delivery in real-world settings [24]. The real-world evidence generated from this study may be used to inform decision-making on HCRU management for this patient population.

The data source used in the present study differentiates it from prior studies in this area, which have used conventional adjudicated claims-based datasets [12–14,17] and may not accurately represent HCRU and costs in patients with newly diagnosed AML. The PHD includes information on billed services, while adjudicated claims have been evaluated for payment of benefits. In a retrospective claims analysis of the Surveillance, Epidemiology, and End Results-Medicare database (1997–2007), newly diagnosed AML patients accrued $74,177 more in costs than non-AML patients. Hospitalizations accounted for 74% of the total costs [17]. In a retrospective analysis of MarketScan claims databases (2007–2011), patients receiving chemotherapy alone had a mean of four hospitalizations, 52.9 inpatient days, and 52.4 outpatient visits, and those receiving allogeneic hematopoietic cell transplantation (alloHCT) had five hospitalizations, 92.5 inpatient days, and 74.5 outpatient visits during the 12 months after an AML diagnosis. Adjusted mean annual costs of inpatient and outpatient visit-days were $280,788 for patients who received chemotherapy and $544,178 for patients who received alloHCT [13]. In a retrospective analysis of MarketScan claims databases (2009–2015), average in-patient, outpatient, and ED expenditures among patients with AML who were treated at diagnosis and achieved remission were $182,672, $22,430, and $613, respectively, with $20,854 in other outpatient services expenditures [14]. A retrospective observational study using the IMS Health PharMetrics Plus and IMS Hospital Charge Detail Master databases (2007–2016) reported that the mean total costs per commercially insured patient were $386,077 in patients who received treatment (chemotherapy/HSCT), with 60% of total costs ($231,867 per patient) incurred during the initial health state, which represented time without remission/relapse [12].

While informative, the hospital HCRU and costs reported in these prior studies were primarily representative of reimbursements paid to hospitals for commercially insured patients. Our study was designed to provide a more accurate representation of the hospital-based HCRU and costs incurred by patients with newly diagnosed AML who received intensive induction chemotherapy in the USA. Data were derived from the PHD, a large US all-payer, hospital-based healthcare database that provides a complete census of inpatients from geographically diverse hospitals and contains a day of service-stamped log of billed items, including procedures, medications, and laboratory, diagnostic, and therapeutic services at the individual patient level [23]. Study eligibility criteria definitively guided the selection of patients with newly diagnosed AML who received standard of care with standard intensive induction regimens (i.e., 7 + 3 or CPX-351 regimens). HCRU and costs were assessed for variable follow-up following the index date, facilitating a comprehensive understanding of HCRU and costs until remission or relapse. Median age of patients in this study was 56 years, which is consistent with their eligibility for high dose induction chemotherapy [25] but in contrast to statistics that identify the national average age at first diagnosis of AML as 68 years [26].

Findings from our study demonstrated that the majority of patients with newly diagnosed AML who received intensive induction chemotherapy were treated in the urban setting in large hospitals (≥500 bed capacity). Most patients receiving induction therapy in the USA are housed in non-critical care units in academic centers. Previous reports confirm that center type impacts outcomes in patients with AML [27–31]. Data from the Nationwide Inpatient Sample database (2009–2011) showed mortality rates in a matched cohort of patients with AML admitted for chemotherapy (induction or consolidation) at high-volume versus low-volume centers were 1.59% versus 4.07%, respectively [28]. An analysis using the PHD (2010–2017) demonstrated that lower hospital volume compared with high volume was associated with in-hospital death or discharge to a hospice in patients treated with cytarabine-anthracycline-based intensive induction chemotherapy during their first inpatient stay for AML [31]. Data from the National Cancer Database (2003–2011) showed worse 1-month mortality in patients with AML who received their first course of treatment at nonacademic (e.g., community cancer program, comprehensive community cancer program) versus academic centers [27]. The volume-outcome relationship in AML may be attributed to better diagnostics, therapies, and management of complications, and improved healthcare provider experience, adherence to treatment guidelines, and ancillary staff and services at high-volume/academic centers compared with low-volume/community centers [30]. Taken together, these findings demonstrate that newly diagnosed AML is a disease requiring complex management and that patients need access to healthcare resources more readily available at high-volume/academic centers compared with low-volume/community centers.

Similar to previous findings, our study found that newly diagnosed AML is associated with high HCRU and costs, with inpatient stays as the primary cost driver [15,17,18]. As novel, less toxic therapies are developed and improvements in supportive care are introduced, early discharge of select patients after induction chemotherapy and/or the management of induction therapy administration in the outpatient setting may reduce healthcare costs and improve quality of life among patients with newly diagnosed AML receiving induction chemotherapy [32–34]. A pilot study conducted to explore the safety and feasibility of intensive initial or salvage induction chemotherapy administered in the outpatient setting for patients with AML and high-risk myelodysplastic syndrome with no significant organ dysfunction and low risk of early death found that complete outpatient induction chemotherapy is feasible, and a successful implementation requires a multidisciplinary team-based approach [32]. Among patients in our study, high rates of neutropenia, sepsis, pneumonia and fungal infections were identified. These hematologic adverse events can carry an increased probability of hospital readmission and may pose a significant barrier to widespread implementation of early hospital discharge and outpatient induction chemotherapy [35,36].

In our study, patients with newly diagnosed AML who received subsequent HSCT had high HCRU and costs. Implementing maintenance therapy may ultimately lower the overall economic burden of AML. In patients with AML who achieve complete remission following induction therapy and who are not candidates for HSCT, effective, well-tolerated maintenance therapy may reduce the risk of relapse [9]. Notably, maintenance therapy has been associated with lower HCRU than placebo among patients with de novo or secondary AML in first remission who were not candidates for HSCT [37].

This study has several limitations, including those inherent to any observational retrospective study. First, this study relied on accurate and complete diagnosis data, procedure coding, and similar cost accounting methods between hospitals. Clinical details, such as laboratory test results, are not available for all patients in the PHD. Second, diagnosis codes were used to identify AML diagnoses. As there is no specific date associated with a diagnosis code, we used a proxy of the AML diagnosis date based on the admission or discharge date. Patients could have been treated in clinics that were not affiliated with the hospitals; therefore, the actual AML diagnosis date could have differed from what was identified in the hospitals. Third, patients could only be followed if they were treated in the same hospital; therefore, it was not possible to examine previous medical history for the patients if they had been treated in other hospitals or physicians' offices. Fourth, patients may have been lost to follow-up due to switching to a different healthcare provider. Last, the study only captured medications received during inpatient hospital admissions and did not include prescriptions filled in pharmacies.

Conclusion

In the US hospital setting, HCRU and costs for patients with newly diagnosed AML who receive intensive induction chemotherapy are substantial, and primarily driven by lengthy inpatient stays. The majority of patients received treatment at large teaching hospitals in the urban setting, demonstrating the high intensity of care needed by this patient population. Novel treatment options that reduce hospitalization rates among patients newly diagnosed with AML who have received intensive induction chemotherapy may alleviate some of the economic burden of AML.