Diagnostic amyloid light chain amyloidosis hospitalizations associated with high acuity and cost: analysis of the Premier Healthcare Database

This retrospective cohort study used 2017–2020 data from the Premier^® Healthcare Database (Premier, Inc., NC, USA), which contains clinical coding, hospital cost and patient billing data from approximately 25% of all US hospital discharges. The database contains a record of all billed items including medications, laboratory, diagnostic and therapeutic services. Diagnosis fields include admitting and primary diagnosis, with the latter defined by the Uniform Hospital Discharge Data Set as the condition chiefly responsible for the admission. The charge and cost for each hospitalization is available. Charges are amounts billed to the payer. In contrast, costs represent payment for services and include variable expenses (those related directly to the activity of the relevant department such as supplies and direct patient care) and fixed expenses (including depreciation, management, repair and maintenance and overhead). Costs reported by the Premier Healthcare Database are based on a combination of information from individual hospital cost-accounting systems and calculations using Medicare costs to charges ratios. Reported costs do not include professional fees for services by physicians and other independent practitioners [6]. Patients can be tracked across hospitalizations if they are readmitted to the same hospital. Data are at the admission level, and patients admitted more than once are considered separate admissions. The terms admission and patient are used interchangeably for simplicity.

The study population included hospitalized patients aged ≥18 years with ≥1 inpatient claim containing the ICD-10-CM for AL amyloidosis (E85.81) during the study period of 1 October 2017–31 December 2020. This period was selected because the code E85.81 specific for AL amyloidosis was not available before October 2017. Patients were excluded if they had a diagnosis for other types of amyloidosis (ICD-10-CM codes: E85.0x-E85.3x), chronic inflammatory disease (e.g., rheumatoid arthritis [ICD-10-CM: M05.40–M06.9]), inflammatory bowel disease (Crohn’s disease [ICD-10-CM: K50.xx]), ulcerative colitis (ICD-10-CM: K51.xx), bronchiectasis (ICD-10-CM: J47.xx Q33.4) or chronic osteomyelitis (ICD-10-CM: A02.24, H05.02x; M86.xx). When more than one qualifying hospitalization for a patient existed, only the first one was included. We reported data on patient socio-demographics, payer type, comorbidities, discharge status, hospital characteristics, length of stay (LOS), intensive care unit (ICU) use, mortality and hospitalization costs. Patients were stratified into diagnostic and other hospitalization. A diagnostic hospitalization was one where the patient had a code for bone marrow, kidney, liver, abdominal fat pad, salivary gland, gingival or endomyocardial biopsy during the stay. In addition, hospitalizations were not considered diagnostic if they included evidence of a solid organ or hematopoietic cell transplant.

Descriptive statistics including means with standard deviations (SD) and relative frequencies with percentages for continuous and categorical data, respectively, were calculated. To compare between diagnostic versus other hospitalizations, t-test and χ² test were performed for continuous and categorical variables respectively. Costs were inflated to 2020 USD using the medical care component of the Consumer Price Index [7]. For context, we compared costs to the most recent national averages calculated from the 2018 National Inpatient Sample and inflated to 2020 USD. Statistical analyses were performed using SAS^© version 9.4.

Between 2017 and 2020, 1419 hospitalizations included ≥1 inpatient claim for AL amyloidosis. Of these, 60 were excluded for having codes consistent with other amyloidosis subtypes and 18 for having codes for comorbid chronic inflammatory conditions, leaving 1,341 patients in the study sample (Table 1). The mean (SD) age at admission was 67.2 (11.2) years. Male patients made up 65.9% of the sample. Most (64.3%) were White with 22.8% Black, 1.4% Asian and 11.4% other or undetermined. Medicare was the primary payer for 62.4% of admission, followed by managed care at 16.3%, Medicaid at 9.5% and commercial indemnity plans at 7.5%. Admissions were relatively evenly distributed across years 2018–2020 (28.7, 35.0 and 30.4%, respectively) with 6.0% from 2017 (the ICD code for AL amyloidosis became available in October 2017). Most admissions were to teaching hospitals (62.4%) in urban areas (91.2%; results not shown). The mean (SD) (median) LOS was 9.5 (9.7) (6.0) days. During the hospital stay, 20.1% of patients were admitted to the ICU and had a mean (SD) ICU LOS of 6.5 (7.6) days. A total of 107 (8%, 95% CI: 6.5–9.4%) patients died in the hospital. The mean (SD) total hospitalization cost was $27,099 ($34,849) for hospitalized patients with AL amyloidosis. In comparison, the mean cost for all US hospital stays was $14,661 ($129) (based on 2018 data inflated to 2020 USD) [8], and the mean cost for all US hospital stays with a principal diagnosis of cancer was $23,249 (2017 data inflated to 2020) [9].

Of 1341 hospital admissions, 236 (17.6%) were considered diagnostic (Table 1). Patient demographics, payer type, year of admission and teaching hospital status did not differ between diagnostic and other admissions. Bone marrow and kidney biopsies were the most common types of biopsies performed during diagnostic hospitalization (79.5 and 44.9%, respectively), while only five cases of endomyocardial biopsy – typically, an outpatient procedure – were performed among these admissions. Diagnostic hospitalizations had longer LOS (mean [SD] 14.5 [11.7] vs 8.4 [8.9] days; p < 0.001), higher cost ($40,052 vs $24,360; p < 0.001) and higher total charges ($161,526 vs $104,129; p < 0.001) than nondiagnostic ones, but mortality was similar between the two groups (Figure 1).

AL amyloidosis often presents a diagnostic challenge. Patients with nonspecific symptoms like fatigue or symptoms typical of other conditions, like shortness of breath, often have multiple rounds of inconclusive testing over months or years [5]. Our analysis provides several novel insights into hospitalizations among patients with AL amyloidosis. Approximately one in six hospitalizations with AL amyloidosis is associated with the initial diagnosis of the disease. Of diagnostic admissions, one in five spent time in the ICU and one in 13 died before discharge. Diagnostic admissions are more likely to be urgent or emergent, require longer hospital stays and have substantially higher costs compared with hospitalizations in patients with known AL amyloidosis. To better understand the drivers of the cost gap between diagnostic and other admissions, we compared daily costs as well as average costs in various cost categories between these groups. Diagnostic hospitalizations were less costly per day than other hospitalizations ($2775 vs $3164; p = 0.005), while mean costs for different procedures were not significantly different between the two groups. Therefore, varying utilization of diagnostic and therapeutic services and varying LOS are likely to be the main drivers of cost differences. Notably, there are intrinsic differences in utilization among the two groups; diagnostic admissions have biopsy procedures while transplants only happen in other admissions.

Limited data exist on hospital utilization and cost in AL amyloidosis. A 2018 study using commercial insurance claims data from 2007 to 2015 estimated the LOS at 10.2 days and inpatient cost at just under $37,909 per year, but, as there was no ICD code specific for AL amyloidosis available at the time of that study, the authors relied on a more general code plus the use of specific medications to identify individuals with the disease [10]. A 2019 study using the same database as the current study but focusing on cardiac amyloidosis from all subtypes found an average LOS of 8.3 days and a cost of $20,584 [11].

Our study is limited on the reliance for specific biopsies to identify patients having had a diagnostic admission. This could have led to exclusion of patients whose diagnosis was made in the hospital on the basis of other findings and inclusion of patients whose diagnosis was previously known but had biopsies for other reasons. This would be unlikely as the diagnosis of AL amyloidosis relies on tissue confirmation. Some tissue sites (e.g., fat pad) are specific to AL amyloidosis. It would also be unlikely that a patient with confirmed AL amyloidosis would undergo a subsequent bone marrow biopsy during hospitalization. The Premier database, while widely used for research, includes primarily information relevant to payment for services and miscoding is possible. It does not include data from federally funded hospitals (e.g., Veterans Affairs).

In conclusion, for the first time, AL amyloidosis specific codes to study hospitalizations in a large, nationally representative database provide data on diagnostic hospitalizations in the disease. Patients with AL amyloidosis are quite ill, with 20% spending time in the ICU and 8% dying in the hospital. Those diagnosed in the hospital have longer stays and higher costs than other AL amyloidosis hospitalizations. Continued effort is needed to improve diagnostic pathways toward early diagnosis of this multisystem disease.

This analysis was supported by Prothena Biosciences Ltd., Dublin, Ireland, a member of the Prothena Corporation PLC group. TP Quock is an employee of Prothena Biosciences Inc. and holds stock in Prothena Corporation PLC group. A D’Souza is an employee of the Medical College of WI, USA and was paid by Prothena Biosciences Inc. to consult as a subject matter expert. E Chang, K Bognar, MS Broder and MH Tarbox are employees of PHAR, which received funding from Prothena to conduct the research described in this manuscript. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Partnership for Health Analytic Research provided analysis, writing and editorial support, funded by Prothena Biosciences Ltd.

The Premier data are de-identified, compliant with the Health Insurance Portability and Accountability Act (HIPAA) and thus consisted with 45 CFR 46.101(b)(4) and exempt from Institutional Review Board oversight.

This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-nd/4.0/