Healthcare resource utilization and costs in amyloid light-chain amyloidosis: a real-world study using US claims data

We used data from the Truven MarketScan^® Commercial and Medicare Supplement Databases (Truven Health Analytics, MI, USA) from 1 January 2007 to 31 December 2015 to estimate the annual healthcare utilization and costs among AL amyloidosis patients.

These claims data provide a unique opportunity to examine this rare disease because it include a sufficient sample of patients. The MarketScan Commercial and Medicare Supplemental Databases are Health Insurance Portability and Accountability Act-compliant administrative claims databases. The commercial data included medical encounters from approximately 65 million individuals and their dependents insured by employer-sponsored plans (i.e., non-Medicare eligible). Coverage was provided under a variety of fee-for-service, fully capitated and partially capitated health plans, including preferred provider organizations, point of service plans, indemnity plans and health maintenance organizations. The Medicare supplemental data included about 5.3 million Medicare-eligible retired employees and their spouses with employer-sponsored Medicare supplemental plans. Given the deidentified nature of the data used in the present study, informed consent was not required by Health Insurance Portability and Accountability Act rules.

As there is no diagnosis code specific to AL amyloidosis, adults ≥18 years old with AL amyloidosis were identified if they had: at least one inpatient claim or at least two outpatient claims consistent with AL amyloidosis (International Classification of Diseases, 9th Revision, Clinical Modification [ICD-9-CM] codes: 277.30 [amyloidosis, unspecified] or 277.39 [Other amyloidosis]; International Classification of Diseases, 10th Revision, Clinical Modification [ICD-10-CM] codes: E85.4x [organ-limited amyloidosis], E85.8x [other amyloidosis] or E85.9x [amyloidosis, unspecified]) in any diagnosis field; and received one AL-specific treatment (e.g., chemotherapy and HSCT) on or after the first amyloidosis diagnosis in the study period. We identified two separate groups of patients – prevalent and incident-as described below.

For prevalent patients, we reported the following patient demographic, clinical characteristics and all-cause healthcare utilization and costs in each calendar year. For patients with incident AL amyloidosis, we reported the patient demographic and clinical measures during baseline (1-year period prior to the index date) and the healthcare resource utilization and cost measures based on follow-up enrollment lengths. Specifically, for patients with a follow-up length ≤12 months, we reported their actual healthcare utilization and costs without annualization. For a subgroup of patients who had follow-up length ≥24 months, we reported their annualized healthcare utilization and costs in each follow-up year.

In our claims databases, there were 7326 patients identified overall (27.5% [n = 2018] had incident AL amyloidosis), with a mean (SD) age 63.6 (12.1), 45% of whom were female. All US regions were represented, and the majority of patients had commercial insurance (Table 1).

Individuals with AL amyloidosis had a substantial burden of comorbidities or concurrent conditions, as determined by having at least one claim with a relevant ICD-9/ICD-10 code (Table 1). The mean (SD) of CCI in all prevalent patients and incident patients was 4.3 (3.2) and 3.3 (2.9), respectively. Common individual comorbidities in the CCI were renal disease (39.3% in prevalent patients; 26.4% in incident patients), congestive heart failure (33.2; 23.1%) and moderate or severe liver disease (28.6; 25.3%). Other common conditions observed included multiple myeloma (38.9% in prevalent patients; 15.5% in incident patients), monoclonal gammopathy of undetermined significance (19.6; 15.8%) and hypothyroidism (17.7; 16.2%) (Table 1).

The majority (68.9% [n = 5044]) of all patients used AL chemotherapy-based treatment at least once during the 9-year period (Table 2). The use of any AL chemotherapy-based treatment remained stable during the study period. However, for specific treatments, there was an increased use of bortezomib during the study period from 4.6% in 2007 to 25.3% in 2015; in contrast, the use of melphalan decreased from 18.9% in 2007 to 2.0% in 2015.

Hospitalization was common: 50.1% (n = 3670) of the overall study sample was admitted at least once and 11.3% (n = 827) were hospitalized three or more times (Table 2). Among admitted patients, mean (SD) length of stay was 14.7 (19.5) days (not shown in table). Rates of any hospitalization decreased from 57.0% in 2007 to 46.2% in 2015. Almost three-quarters of patients (72.0% [n = 5274]) had an ED visit at least once during the 9-year period (Table 2) and 6.8% (n = 496) had more than three ED visits (not shown in table). ED visits remained stable during the study period. Patients had a mean (SD) of 45.2 (36.2) non-ED outpatient visits per year. Non-ED outpatient visit rates increased from 30.0% in 2007 to 37.9% in 2015 during the study period.

Mean (SD) total annual all-cause healthcare costs were $101,855 (148,965) for all patients (Table 3). The majority of the total costs were for medical services (mean [SD] $88,801 [144,491]), with $50,126 (86,232) accrued in the outpatient, $37,909 (100,892) inpatient and $766 (3650) ED settings. Total costs increased over time from $92,866 in 2007 to $114,030 in 2015. Total medical (non-outpatient pharmacy) costs increased from $80,099 in 2007 to $97,513 in 2015. ED visit costs increased from $685 in 2017 to $1,043 in 2015. Other outpatient medical costs increased from $36,235 in 2007 to $57,227 in 2015. Outpatient pharmacy costs increased slightly from $12,768 in 2007 to $16,517 in 2015.

In subgroup analyses with incident AL amyloidosis patients, healthcare utilization differed between all incident patients and those with ≥24 months of follow-up (Table 4). With regard to inpatient hospitalizations, among all incident AL amyloidosis patients, 64.6% (n = 1303) were hospitalized at least once during post year 1, and 16.6% (335) were hospitalized three or more times. In comparison, 61.2% (n = 543) of the incident patients with ≥24 months of follow-up were hospitalized at least once during post year 1 and 10.3% (n = 92) were hospitalized three or more times. These rates (among newly diagnosed patients with ≥24 months of follow-up) declined over time, as 31.9% (n = 283) were hospitalized at least once during post year 2 and 5.3% (n = 47) were hospitalized three or more times.

With regard to ED visits, 37.9% (n = 764) of all incident patients had at least one ED visit in post year 1 and 7.4% (n = 150) had at least three ED visits. Among incident patients with ≥24 months of follow-up, 34.9% (n = 310) had at least one ED visit in post year 1 and 31.6% (n = 280) had at least one ED visit in post year 2. Non-ED outpatient service visits declined slightly among patients with ≥24 months of follow-up from a mean (SD) of 49.2 (35.6) visits in post year 1 to 39.7 (36.1) in post year 2.

Mean (SD) total annual all-cause healthcare costs among all incident patients were $122,180 (159,074) in post year 1 (Table 5). Medical (non-outpatient pharmacy) costs were $110,519 (154,625), with $55,679 (77,855) accrued in the outpatient, $54,053 (116,851) inpatient and $787 (4082) ED settings. Among incident patients with ≥24 months of follow-up, total mean (SD) costs decreased over time from $104,672 (134,993) in post year 1 to $68,502 (121,711) in post year 2 (Table 5). Total medical (non-outpatient pharmacy) costs declined from $92,685 (128,630) in post year 1 to $55,871 (114,643) in post year 2.

In our analysis with all prevalent patients, the use of any chemotherapy treatment remained stable during the study period, but specific treatments, such as bortezomib and melphalan increased and decreased, respectively, from 2007 to 2015. This change in treatment practice over the 9-year study period is likely due to the fact that: studies published in this time period showed that bortezomib, with its rapid action and ability to inhibit proteasome enzyme complexes within cells, leads to high response rates and good tolerance; [19,20] and bortezomib had been increasingly used as a first-line therapy in AL amyloidosis during the study period. Hospitalization was common: half of AL amyloidosis patients were admitted at least once during the year and more than 11% were hospitalized at least three-times.

The average total annual healthcare costs among all prevalent AL amyloidosis patients from 2007 to 2015 data were $101,855. These costs, which included outpatient pharmacy and medical costs (i.e., inpatient hospitalizations, ED visits and outpatient medical costs), increased over time from $92,866 in 2007 to $114,030 in 2015. Because our analysis is the first known study to examine healthcare costs over time associated with AL amyloidosis, we are unable to compare this trend to other analyses with this specific patient population.

Among incident AL amyloidosis patients, in the first year after diagnosis, healthcare utilization and costs were lower in a subgroup of patients who had at least 2 years of follow-up, compared with patients with shorter post-index follow-up enrollment. This may be because patients who lived longer (i.e., at least 2 years post-diagnosis) had less severe disease and/or were healthier overall. For example, these patients with at least 2 years of follow-up had a slightly lower burden of comorbidities, as measured by the CCI, compared with patients with less than 2 years of follow-up.

Additionally, in a subgroup of patients with at least 2 years of follow-up, healthcare utilization and costs declined over time. Specifically, rates of hospitalization declined from 61.2% 1 year after diagnosis to 31.9% 2 years after diagnosis. Also, rates of AL chemotherapy-based treatment declined slightly from 67.5% 1 year after diagnosis to 54.8% in the second year after diagnosis. Total costs declined from $104,672 in the first year after diagnosis to $68,502 in the second year after diagnosis. Similar trends have been shown in breast, lung, colorectal and prostate cancer [21,22]. Although we did not examine diagnostic tests or surgeries specifically, it may be that more of these tests and/or surgeries are performed in the first year after diagnosis, with more of them likely to be from inpatient facilities, than later in the disease course, leading to higher costs immediately after diagnosis.

This study does not shed light on long-term healthcare utilization and costs among the incident patients. If we had been able to follow all of these newly diagnosed AL amyloidosis patients until death, we likely would have found an increase in costs over their lifetime, peaking in the final months of life. This hypothesis is supported by a previous study that found the greatest costs to be accrued at the time of initial diagnosis and in the final months of life among heart failure patients [23]. In other words, when these newly diagnosed AL amyloidosis patients are examined as part of the prevalent population, their healthcare costs increase over time, as shown in our analysis.

This study has limitations. First, there is no diagnosis code in ICD-9-CM or ICD-10-CM specific to AL amyloidosis and no generally accepted or clinically validated method for identifying this condition using health insurance claims data, the codes for this study were selected with clinical expert input to eliminate as many non-AL amyloid patients as possible (e.g., by excluding 277.31, familial Mediterranean fever and E85.3, secondary systemic amyloidosis). We further required patients to have received treatment consistent with expert recommendations for AL amyloidosis. This requirement would be expected to decrease the sensitivity but increase the specificity of our identification algorithm. Nonetheless, a small proportion of patients with transthyretin-related hereditary amyloidosis would likely still have been included in our sample. We are planning to validate the algorithm we used in a future study using data collected from medical records. Second, our dataset included commercially insured patients only who were either healthy enough to be employed or closely related to an employed person, so findings may not precisely reflect the US population due to this healthy-worker bias. Third, the administrative claims used in this study were collected for reimbursement purposes and the completeness and accuracy of medical coding is subject to data coding restrictions and data entry error. Fourth, presence of comorbidities was determined based upon having at least one claim with a relevant ICD-9/ICD-10 code, not a clinical diagnosis, so misclassification, diagnostic uncertainty or coding errors were possible. Fifth, we observed a high overlap of multiple myeloma and AL amyloidosis in prevalent patients. Our data are not clinically detailed enough to determine which diagnosis was most appropriate for these patients. Sixth, the healthcare cost estimates may be underestimated as claims for patients who die outside of the hospital were not available, and services not covered by insurance or rendered ‘out-of-network’ were not included. Seventh, the subgroup of patients with at least 2-year enrollment were less sick, as measured by CCI and had lower comorbid disease burden, compared with those with shorter enrollment. For example, in patients with less than 1 year of enrollment, 31.4% of them had congestive heart failure, compared with 17.4% found in patients with at least 2 years of enrollment. This reduces the generalizability of the cost findings in this healthier subgroup. Lastly, cost estimates include direct healthcare costs only, and do not take into account important indirect costs associated with caregiver burden, loss of productivity or reduced quality of life.