Temporal trends and disparities in cardiovascular disease-related mortality among adults with multiple myeloma in the United States, 1999–2023

Abstract

Background

Novel therapeutic agents have substantially improved multiple myeloma (MM) survival, making cardiovascular disease (CVD) a predominant cause of non-cancer mortality. Because MM predominantly affects older adults (median age 69 years) at elevated cardiovascular risk, we focused on adults aged ≥ 45 years. Comprehensive national data on temporal trends and demographic disparities in CVD-related mortality among MM patients remain limited.

Methods

We conducted a retrospective analysis using CDC WONDER Multiple Cause-of-Death database data (1999–2023) for U.S. adults aged ≥ 45 years with MM and CVD as underlying or contributing causes of death. Age-adjusted mortality rates (AAMRs) per 100,000 population were standardized to the 2000 U.S. standard population using the direct method. Joinpoint regression analysis identified temporal changepoints and calculated annual percent changes (APCs) and average annual percent changes (AAPCs).

Results

Among 117,907 CVD-related deaths in MM patients, overall AAMR was 3.92 per 100,000 (95% CI: 3.80–4.03) with no significant long-term change (AAPC + 0.04%). Joinpoint analysis revealed five phases: non-significant increase from 1999 to 2001 (APC + 3.10%), significant decline from 2001 to 2009 (APC − 2.05%), stability from 2009 to 2018 (APC + 0.16%), marked increase from 2018 to 2021 (APC + 6.70%, P < 0.05), and significant decline from 2021 to 2023 (APC − 4.59%). Males had significantly higher mortality than females (5.23 vs. 2.98 per 100,000; P < 0.001). Non-Hispanic Black adults experienced highest burden (8.38 per 100,000), 2.4-fold higher than non-Hispanic White adults (3.51 per 100,000). Mortality increased significantly with age (P < 0.001 for trend), reaching 18.46 per 100,000 in adults ≥ 85 years. Hypertensive diseases (AAPC + 4.79%) and cardiac arrhythmias (AAPC + 3.71%) showed largest increases, while ischemic heart diseases (AAPC − 1.15%) and cerebrovascular diseases (AAPC − 1.07%) declined. Geographic analysis revealed the highest mortality rates in the Northeast (4.29 per 100,000) and West (4.14 per 100,000).

Conclusions

Although overall CVD mortality remained stable over 25 years, concerning increases in hypertensive, arrhythmic, thromboembolic, and heart failure mortality during 2018–2021 coincided with expanded cardiotoxic therapy use and COVID-19 disruptions. These findings highlight the importance of incorporating baseline cardiovascular risk assessment and longitudinal surveillance into MM treatment pathways, ensuring adherence to guideline-based thromboprophylaxis, and developing multidisciplinary cardio-oncology models that address persistent racial and ethnic disparities in cardiovascular outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12872-025-05417-w.

Introduction

Multiple myeloma (MM) accounts for approximately 10% of hematological malignancies and predominantly affects older adults, with a median age at diagnosis of 69 years [1, 2]. The therapeutic landscape has undergone revolutionary changes over the past two decades with the introduction of novel agents, including proteasome inhibitors (PIs), immunomodulatory drugs (IMiDs), and monoclonal antibodies [3, 4]. These advances have dramatically improved survival outcomes, extending median overall survival from 3 years to 8–10 years [5, 6]. Consequently, an expanding population of MM survivors faces long-term treatment sequelae and age-related comorbidities, shifting clinical focus toward comprehensive survivorship care. Among these concerns, cardiovascular disease (CVD) has emerged as a critical priority [7, 8].

Population-based studies demonstrate that cardiovascular complications are common among MM survivors, with incidence rates ranging from 20% to over 40% depending on cohort characteristics and follow-up duration [9, 10]. CVD accounts for approximately 5–12% of all deaths [7, 11] and represents a leading non-cancer cause of mortality, reflecting complex interplay among patient demographics, disease biology, and treatment-related factors [7, 12]. While mechanistic understanding of cardiovascular toxicity has advanced considerably, population-level temporal patterns remain poorly characterized, creating a critical need for comprehensive epidemiologic surveillance to inform clinical practice and public health policy.

The pathways underlying increased CVD risk are multifaceted. Therapy-related cardiotoxicity, particularly from PIs, constitutes a major contributor. Carfilzomib exhibits pronounced cardiovascular toxicity, with meta-analyses reporting all-grade cardiovascular adverse events in approximately 18.1% of patients (grade ≥ 3 in 8.2%), manifesting as heart failure, hypertension, arrhythmias, and ischemic heart disease (IHD) [13, 14]. IMiD-driven thromboinflammation represents another critical pathway, conferring increased venous thromboembolism (VTE) risk that is markedly amplified when combined with high-dose corticosteroids or multiagent chemotherapy, with VTE incidence reaching up to 26% in some lenalidomide plus high-dose dexamethasone cohorts [15, 16]. Contemporary cardio-oncology guidelines emphasize structured risk assessment, surveillance, and multidisciplinary management across the cancer-care continuum [17, 18].

Despite the recognized importance of CVD in patients with MM, critical knowledge gaps persist regarding cardiovascular mortality trends and disparities. Prior investigations have been constrained by single-center designs, pre-novel-agent populations, focus on specific treatment regimens or short-term outcomes, or absence of comprehensive longitudinal evaluation [19]. Few studies have systematically quantified national temporal trends in CVD-related mortality among MM patients over extended periods using sophisticated statistical approaches such as joinpoint regression [8]. No prior nationwide analysis has extended through the COVID-19 pandemic era or provided comprehensive demographic and geographic stratification across a 25-year timespan. The temporal evolution of CVD mortality risk during the late 2010 s and early 2020 s remains poorly characterized, particularly regarding newer therapeutic agents and the pandemic’s impact on cancer care delivery and cardiovascular outcomes [20]. Additionally, existing literature lacks granular analysis of specific CVD subtypes and their distinct temporal patterns, limiting understanding of which cardiovascular conditions drive mortality trends and hampering the development of targeted prevention strategies.

This study aimed to characterize temporal trends and population disparities in CVD mortality among U.S. adults aged ≥ 45 years with MM from 1999 to 2023, using national death certificate data and joinpoint regression analysis. This investigation addresses critical knowledge gaps by providing nationwide, contemporary estimates spanning 25 years with granular subgroup and geographic resolution. We hypothesized that temporal increases in specific cardiovascular mortality subtypes—particularly hypertensive disease, cardiac arrhythmias, heart failure, and thromboembolic events—would coincide with expanded use of cardiotoxic therapies (PIs and IMiDs) and healthcare disruptions during the COVID-19 pandemic, while traditional atherosclerotic endpoints (IHDs and cerebrovascular disease) would continue long-term declining trends observed in the general population. These findings will inform cardio-oncology risk stratification strategies, surveillance protocols for high-risk populations, and evidence-based policy development for improving cardiovascular outcomes in MM survivors.

Methods

Study design and data source

We conducted a retrospective, population-based analysis of mortality data from the Centers for Disease Control and Prevention (CDC) Wide-Ranging Online Data for Epidemiologic Research (WONDER) Multiple Cause-of-Death database covering 1999–2023 [21]. This publicly available dataset contains all death certificates from the 50 United States and District of Columbia, providing underlying and contributing causes of death coded according to the International Classification of Diseases, Tenth Revision (ICD-10), along with demographic and geographic information. The analysis covered deaths occurring between January 1, 1999, and December 31, 2023, and data were accessed on August 15, 2025. Detailed query parameters—including ICD-10 codes, demographic filters, and variable groupings—are provided in the Supplemental Methods.

Study population

We included decedents aged ≥ 45 years with MM (ICD-10 code C90.0) and CVD (I00–I99) listed anywhere on the death certificate as underlying or contributing cause of death. Individuals with secondary plasma cell neoplasms or concurrent non-myeloma malignancies were excluded to ensure diagnostic specificity. The age threshold of ≥ 45 years was selected to capture the population most affected by MM and to ensure adequate sample sizes for subgroup analyses, as approximately 95% of MM cases occur in individuals aged ≥ 45 years, with a median age at diagnosis of 69 years.

CVDs were further categorized into six subtypes based on ICD-10 codes: (1) IHDs (I20-I25), including angina pectoris, myocardial infarction, and other acute and chronic ischemic conditions; (2) hypertensive diseases (I10-I15), including essential hypertension, hypertensive heart disease with or without heart failure, hypertensive renal disease with or without renal failure, combined hypertensive heart and renal disease, and secondary hypertension; (3) cerebrovascular diseases (I60-I69), including subarachnoid hemorrhage, intracerebral and other intracranial hemorrhages, cerebral infarction, unspecified stroke, and other cerebrovascular conditions; (4) heart failure (I50), including congestive heart failure, left ventricular failure, systolic and diastolic heart failure; (5) cardiac arrhythmias (I47-I49), including paroxysmal tachycardia, atrial fibrillation and flutter, and other cardiac arrhythmias; and (6) thromboembolic diseases (I26, I80-I82), including pulmonary embolism, phlebitis and thrombophlebitis, and other venous embolism and thrombosis. Decedents could have multiple CVD codes recorded on death certificates; each individual was counted once in the overall CVD mortality analysis but could contribute to multiple CVD subtype categories when applicable, allowing assessment of the full spectrum of cardiovascular pathology in this population.

Demographic and geographic variables

Race and ethnicity were classified according to U.S. Office of Management and Budget standards as non-Hispanic (NH) White, NH Black or African American, NH Other (American Indian/Alaska Native and Asian/Pacific Islander), and Hispanic/Latino [22]. Annual population denominators stratified by sex, race, and ethnicity were obtained from U.S. Census Bureau intercensal and postcensal estimates [23].

Geographic variables included urbanization status using the 2013 National Center for Health Statistics Urban–Rural Classification: large metropolitan areas (≥ 1,000,000 population), medium/small metropolitan areas (50,000–999,999 population), and rural areas (< 50,000 population) [24]. Regional classification used the four U.S. Census regions: Northeast, Midwest, South, and West [25]. State-level data with fewer than 10 deaths were suppressed by CDC WONDER in accordance with confidentiality standards and excluded from geographic rate calculations to ensure statistical reliability.

Statistical analysis

Crude mortality rates (CMRs) and age-adjusted mortality rates (AAMRs) were calculated per 100,000 U.S. population, with 95% confidence intervals. The overall AAMR represented the pooled age-adjusted mortality rate across the entire study period, computed by direct standardization to the 2000 U.S. standard population. Annual AAMRs for Joinpoint regression were calculated using the same standardization method, consistent with CDC WONDER protocols. All subgroup-specific mortality rates (sex, race/ethnicity, and geography) were age-adjusted to ensure comparability by eliminating confounding from differences in age structure. Age-specific mortality rates were reported as crude rates within each stratum, as age adjustment is not applicable within homogeneous age groups.

Temporal trends were assessed using Joinpoint regression analysis (Joinpoint Regression Program, version 5.1.0.0; National Cancer Institute), which identified statistically significant changepoints in mortality trends and estimated annual percent changes (APCs) for each segment and average annual percent changes (AAPCs) over the entire study [26, 27]. The analysis applied a log-linear model to annual AAMRs, with the optimal number of joinpoints determined via Monte Carlo permutation testing (α = 0.05, 4,499 permutations) using a grid-search algorithm. Given the 25-year observation period, up to four joinpoints (five segments) were permitted, with a minimum segment length of two years to capture pandemic-era fluctuations while maintaining temporal stability. Model selection employed the weighted Bayesian Information Criterion (wBIC), with the final four-joinpoint model selected based on permutation significance (P < 0.001).

APCs represented the estimated annual percentage change in mortality during each segment, assuming a log-linear relationship between calendar year and log-transformed mortality rate. AAPCs summarized the weighted average of segment-specific APCs, with weights proportional to segment duration. Subgroup analyses with suppressed data (< 10 deaths) were excluded from Joinpoint regression to maintain reliability. Parallelism tests were applied to assess whether temporal trends differed significantly between subgroups; significant interaction p-values indicated heterogeneous AAPCs across groups.

Sensitivity analyses were performed to evaluate potential pandemic-period effects on temporal trends. Three analytical scenarios were examined: (1) the full study period (1999–2023), (2) pre-pandemic period only (1999–2019), and (3) excluding pandemic years (1999–2023, excluding 2020–2021). For each scenario, we estimated the annual percent change (APC) using ordinary least squares (OLS) linear regression of annual AAMRs on calendar year, with APC approximated as (e^slope − 1) ×100.

Analyses were conducted using Joinpoint software (version 5.1.0.0) [28] and R software (version 4.4.2; R Foundation for Statistical Computing, Vienna, Austria). Two-sided P < 0.05 was considered statistically significant.

Given the exploratory nature of subgroup analyses, no adjustment for multiple comparisons was applied to avoid excessive type II error. However, this approach may increases the risk of false-positive findings due to multiple statistical tests across numerous demographic strata, geographic divisions, and CVD subtypes. Therefore, subgroup findings are interpreted as hypothesis-generating, emphasizing effect sizes, confidence intervals, and consistent patterns across related subgroups rather than isolated p-values.

Ethical considerations

Institutional review board approval and informed consent were not required because the CDC WONDER database contains publicly available, de-identified data. All analyses complied with CDC data-use policies for public datasets and followed STROBE guidelines for observational research.

Results

Overall mortality trends

During 1999–2023, 117,907 CVD-related deaths occurred among U.S. adults aged ≥ 45 years with MM, representing 42.0% of all MM-related deaths (N = 280,708) during this period. Over the 25-year period, the overall AAMR was 3.92 per 100,000 (95% CI: 3.80–4.03), with no significant long-term change (AAPC + 0.04%; 95% CI: −0.22% to + 0.31%). However, joinpoint analysis revealed five distinct temporal phases masking this apparent stability: early increase (1999–2001), sustained decline (2001–2009), prolonged plateau (2009–2018), marked pandemic-era rise (2018–2021, APC + 6.70%, P < 0.05), and recent decline (2021–2023, APC − 4.59%, P < 0.05) (Table 1; Fig. 1; Supplemental Table 1).

Table 1.

Distribution and trends of cardiovascular disease (CVD)-related mortality among adults with multiple myeloma (MM) in the united States, 1999–2023

Characteristic	Deaths (n)	AAMR per 100,000 (95% CI)	AAPC (95% CI)
Overall	117,907	3.92 (3.80, 4.03)	0.04 (−0.22, 0.31)
Sex
Male	66,410	5.23 (5.02, 5.43)	0.18 (0.02, 0.35)
Female	51,497	2.98 (2.85, 3.11)	−0.40 (−0.81, −0.01)
Age group (years)
45–54	4,216	0.40 (0.34, 0.46)	−0.74 (−1.36, −0.13)
55–64	14,054	1.60 (1.47, 1.73)	−1.11 (−1.38, −0.85)
65–74	30,916	5.10 (4.81, 5.39)	−0.69 (−1.13, −0.22)
75–84	43,041	12.24 (11.66, 12.82)	−0.12 (−0.39, 0.10)
85+	25,680	18.46 (17.32, 19.60)	1.25 (0.96, 1.55)
Race/ethnicity
NH White	83,580	3.51 (3.39, 3.64)	0.16 (−0.11, 0.45)
NH Black	22,658	8.38 (7.81, 8.94)	−0.35 (−0.94, 0.52)
Hispanic/Latino	8,152	3.57 (3.15, 3.99)	−0.52 (−1.03, 0.02)
NH Other (AI/AN, Asian/PI)	3,294	2.45 (2.00, 2.92)	−1.01 (−1.57, −0.28)
CVD subtype
Ischaemic heart diseases	29,646	1.00 (0.94, 1.05)	−1.15 (−1.66, −0.72)
Hypertensive diseases	31,130	1.00 (0.95, 1.06)	4.79 (4.22, 5.65)
Cerebrovascular diseases	11,608	0.39 (0.35, 0.43)	−1.07 (−1.65, −0.49)
Heart failure	25,933	0.86 (0.81, 0.91)	0.80 (0.49, 1.14)
Cardiac arrhythmias	16,985	0.54 (0.50, 0.58)	3.71 (3.05, 4.44)
Thromboembolic diseases	5,464	0.16 (0.14, 0.19)	1.79 (1.04, 2.74)
Geographic region
Northeast	25,015	4.29 (4.02, 4.56)	−1.18 (−1.55, −0.81)
Midwest	24,249	3.62 (3.39, 3.85)	−0.15 (−0.58, 0.22)
South	41,813	3.76 (3.58, 3.95)	0.32 (−0.04, 0.64)
West	26,830	4.14 (3.88, 4.39)	0.34 (0.05, 0.64)
Urban–rural status
Large metropolitan	52,178	4.01 (3.85, 4.17)	−0.18 (−0.70, 0.15)
Medium/small metropolitan	29,581	3.70 (3.50, 3.90)	0.21 (−0.17, 0.55)
Rural	17,164	3.78 (3.51, 4.05)	0.04 (−0.25, 0.35)

Abbreviations: AAMR Age-adjusted mortality rate, AAPC Average annual percent change, CI Confidence interval, CVD Cardiovascular disease, MM Multiple myeloma, NH Non-Hispanic, AI/AN American Indian/Alaska Native, PI Pacific Islander

Note: Death counts may not sum to totals because of overlapping categories. Urban-rural classification is based on the NCHS Urban-Rural Classification Scheme for Counties

Fig. 1.

Overall and sex-specific cardiovascular disease mortality trends in adults aged ≥ 45 years with multiple myeloma, United States, 1999–2023. Age-adjusted cardiovascular mortality rates exhibited five distinct temporal phases: early increase (1999–2001), sustained decline (2001–2009), prolonged plateau (2009–2018), marked pandemic-era rise (2018–2021, APC + 6.70%, P < 0.05), and recent decline (2021–2023, APC − 4.59%, P < 0.05). Males consistently demonstrated higher mortality rates than females, with a modest overall increase (AAPC + 0.18%), whereas female mortality declined significantly (AAPC − 0.40%, P < 0.05). Joinpoint inflection years occurred at 2001, 2009, 2018, and 2021 Note: APC, annual percent change; AAPC, average annual percent change. Rates per 100,000, age-adjusted to the 2000 U.S. standard population; * indicates P < 0.05 by Joinpoint

Sex-specific mortality trends

Males experienced significantly higher mortality rates than females throughout the study period and demonstrated modestly increasing trends (AAPC + 0.18%), while female mortality declined significantly (AAPC − 0.40%; Table 1). Temporal patterns differed substantially by sex (Fig. 1; Supplemental Table 2). Males exhibited volatility with five distinct phases, including pronounced increases during 2018–2021 followed by sharp recent decline. In contrast, females demonstrated a simpler biphasic pattern: sustained decline through 2014 followed by reversal to increasing trends thereafter (Supplemental Table 1).

Age-specific mortality trends

CVD mortality among MM patients increased progressively with advancing age, with rates rising steeply from the youngest (45–54 years) to oldest (≥ 85 years) age groups (Fig. 2; Table 1; Supplemental Table 3). Temporal trends varied substantially across age groups. The youngest adults (45–54 years) showed a persistent decline over the study period. Middle-aged groups (55–74 years) demonstrated early decreases followed by mid-to-late-2010s rebounds and a modest decline in recent years. Older adults (75–84 years) exhibited a similar biphasic pattern with a transient rise around the late 2010 s to early 2020s. In contrast, the oldest group (≥ 85 years) experienced a sustained increase since the mid-2010s, contributing to an overall upward trend. Joinpoint analyses identified major inflection points clustered around 2010–2015 across most strata.

Fig. 2.

Age-specific cardiovascular disease mortality trends in adults aged ≥ 45 years with multiple myeloma, United States, 1999–2023. Crude mortality rates increased sharply with age—0.40 (45–54 y), 1.60 (55–64 y), 5.10 (65–74 y), 12.24 (75–84 y), and 18.46 (≥ 85 y) per 100,000. Temporal trajectories varied across age strata. The youngest group (45–54 y) showed a sustained decline (APC − 0.74%, P < 0.05). Middle-aged adults (55–74 y) demonstrated early decreases followed by mid-2010s rebounds and a modest decline after 2021. Older adults (75–84 y) exhibited a biphasic pattern with a sharp rise during 2018–2021 (APC + 5.98%, P < 0.05) and subsequent decline (APC − 5.36%, P < 0.05). The oldest group (≥ 85 y) showed a sustained increase from 2015–2023 (APC + 4.65%, P < 0.05), resulting in an overall upward trend (AAPC + 1.25%, 95% CI 0.96–1.55). Major joinpoint inflections clustered around the mid-2010s across most age groups Note: APC, annual percent change; AAPC, average annual percent change. Rates per 100,000, age-adjusted to the 2000 U.S. standard population; * indicates P < 0.05 by Joinpoint

Race/ethnicity-specific mortality trends

CVD mortality varied substantially across racial and ethnic groups. NH Black adults experienced the highest burden, with rates approximately 2.4-fold higher than NH White adults. Hispanic/Latino adults demonstrated similar rates to NH White adults, while NH Other adults had the lowest rates (Table 1; Supplemental Table 4). Only NH Other adults showed significant long-term decline (AAPC − 1.01%), while other groups demonstrated non-significant long-term changes.

Temporal patterns generally reflected overall study trends of pre-2010s decline, late-2010s increases, and post-2021 stabilization, though timing and statistical significance varied by group (Fig. 3; Supplemental Table 1). NH White adults exhibited the most complex pattern with five temporal segments mirroring overall trends. Hispanic/Latino adults experienced biphasic decline through 2009 followed by sustained increases. NH Black adults showed similar patterns though without statistical significance, while NH Other adults demonstrated sustained decline throughout.

Fig. 3.

Race/ethnicity-specific cardiovascular disease mortality trends in adults aged ≥ 45 years with multiple myeloma, United States, 1999–2023. Age-adjusted mortality rates (AAMRs) varied markedly by race/ethnicity: non-Hispanic (NH) Black adults had the highest rates (8.38 per 100,000), about 2.4-fold higher than NH White adults (3.51 per 100,000). Hispanic/Latino adults showed comparable rates to NH Whites, while NH Other adults (AI/AN and Asian/PI) had the lowest (2.45 per 100,000) Temporal patterns differed across groups. NH White adults showed five segments with early decline (2001–2009, APC − 2.06%, P < 0.05), plateau (2009–2018), pandemic-era rise (2018–2021, APC + 7.09%, P < 0.05), and post-2021 decline (APC − 4.26%, P < 0.05). Hispanic/Latino adults declined early (1999–2009, APC − 2.52%, P < 0.05) then increased steadily (2009–2023, APC + 0.93%, P < 0.05). NH Black adults had persistently elevated but nonsignificant fluctuations, while NH Other adults showed a consistent decline throughout (APC − 1.01%, P < 0.05). Major inflection points clustered around 2009 and 2018 Note: APC, annual percent change; AAPC, average annual percent change; NH, non-Hispanic.Rates per 100,000, age-adjusted to the 2000 U.S. standard population; * indicates P < 0.05 by Joinpoint

CVD subtype-specific mortality trends

Hypertensive diseases and IHDs accounted for the highest mortality burden among CVD subtypes, each with AAMRs of 1.00 per 100,000 population (Table 1; Supplemental Table 5). Over the study period, four subtypes showed significant increases: hypertensive diseases (AAPC + 4.79%, P < 0.05), cardiac arrhythmias (AAPC + 3.71%, P < 0.05), thromboembolic diseases (AAPC + 1.79%, P < 0.05), and heart failure (AAPC + 0.80%, P < 0.05). Conversely, IHDs (AAPC − 1.15%, P < 0.05) and cerebrovascular diseases (AAPC − 1.07%, P < 0.05) demonstrated significant declines.

Joinpoint analysis revealed distinct temporal patterns for each subtype (Fig. 4; Supplemental Table 1). IHDs declined through 2015 before stabilizing. Hypertensive diseases exhibited pronounced variability with dramatic increases during 2018–2021 and recent stabilization. Cerebrovascular diseases and heart failure demonstrated biphasic patterns with early decline followed by sustained increases from the mid-2010s onward. Cardiac arrhythmias increased markedly after 2013, while thromboembolic diseases increased consistently throughout the study period.

Fig. 4.

Cardiovascular disease subtype-specific mortality trends in adults aged ≥ 45 years with multiple myeloma, United States, 1999–2023. Ischemic heart diseases declined through ~ 2015 before stabilizing; hypertensive diseases showed a large surge during 2018–2021 (segment APC + 14.62%, P < 0.05) with recent attenuation; cerebrovascular deaths fell through 2014 then rose thereafter (2014–2023 segment APC + 3.89%, P < 0.05); heart-failure deaths declined until 2012 then increased (2012–2023 segment APC + 3.72%, P < 0.05); cardiac arrhythmias rose markedly after 2013 (2013–2023 APC + 6.12%, P < 0.05); thromboembolic diseases increased steadily across the period. Major inflection years cluster in the early-to-mid 2010 s and 2018 Note: APC, annual percent change; AAPC, average annual percent change. Rates per 100,000, age-adjusted to the 2000 U.S. standard population; * indicates P < 0.05 by Joinpoint

Geographic region-specific mortality trends

Regional analysis revealed substantial disparities, with the Northeast and West demonstrating the highest mortality rates, followed by the South and Midwest (Table 1; Supplemental Table 6). In absolute terms, the South contributed the most deaths (35.5%), followed by the West (22.8%), Northeast (21.2%), and Midwest (20.6%).

All regions followed similar temporal patterns characterized by early decline, late-2010s increases, and post-2021 stabilization, though magnitude and timing varied (Fig. 5; Supplemental Table 1). The Northeast was the only region demonstrating sustained decline throughout the study period (APC − 1.18%). Other regions exhibited biphasic patterns with early decline followed by recent increases, most pronounced in the South and Midwest during 2017–2021.

Fig. 5.

Age-adjusted cardiovascular disease mortality by U.S. Census region in adults aged ≥ 45 years with multiple myeloma, United States, 1999–2023. The Northeast showed a sustained decline across the period (APC − 1.18%, P < 0.05). The Midwest and South each declined until the mid-/late-2010s then experienced marked rises (Midwest 2017–2021 APC + 6.51%; South 2017–2021 APC + 8.10%, P < 0.05) with attenuation after 2021. The West declined early and then rose gradually from 2013 onward (2013–2023 APC + 1.81%, P < 0.05). Major inflection points cluster in the mid-2010s and around 2017–2018 Note: APC, annual percent change; AAPC, average annual percent change. Rates per 100,000, age-adjusted to the 2000 U.S. standard population; * indicates P < 0.05 by Joinpoint

Urban-rural mortality trends

CVD mortality showed modest variation across urbanization levels, with large metropolitan areas demonstrating the highest rates, followed by rural areas and medium/small metropolitan areas (Table 1; Supplemental Table 7). Long-term trends remained stable across all strata. However, all urbanization levels exhibited consistent temporal patterns characterized by pre-mid-2010s decline followed by late-2010s increases (Fig. 6; Supplemental Table 1).

Fig. 6.

Age-adjusted cardiovascular disease mortality by urban–rural category in adults aged ≥ 45 years with multiple myeloma, United States, 1999–2023. Large metropolitan areas declined through 1999–2018 (APC − 1.06%, P < 0.05) then rose sharply during 2018–2020 (APC + 8.63%, P < 0.05); medium/small metros declined until 2015 (APC − 0.91%, P < 0.05) with increases in 2015–2020 (APC + 3.90%, P < 0.05); rural counties showed a similar pattern (1999–2015 APC − 1.11%, P < 0.05; 2015–2020 APC + 3.80%, P < 0.05). Major inflection years cluster in the mid-2010s Note: APC, annual percent change; AAPC, average annual percent change. Rates per 100,000, age-adjusted to the 2000 U.S. standard population; * indicates P < 0.05 by Joinpoint

State-level mortality trends

State-level analysis revealed substantial interstate variation in both mortality burden and temporal patterns (Fig. 7; Supplemental Table 8). AAMRs typically ranged from 2 to 6 per 100,000 population, with greater variability in small-population jurisdictions where data suppression occurred.

Fig. 7.

State-level death counts (A) and age-adjusted mortality rates (B) for cardiovascular disease in adults aged ≥ 45 years with multiple myeloma, United States, 2023. Note: AAMR, age-adjusted mortality rate. Rates per 100,000, age-adjusted to the 2000 U.S. standard population. States with suppressed counts are labeled NA

Several states demonstrated pronounced late-2010s increases with statistically significant trends, including Illinois, South Carolina, Oklahoma, Maryland, Minnesota, Wisconsin, Virginia, and Mississippi (Supplemental Table 1). Conversely, several states maintained long-term declines throughout 1999–2023, including New York, Connecticut, Missouri, West Virginia, Alabama, and Ohio.

Sensitivity analysis for pandemic-period effects

To assess whether recent increases were influenced by the COVID-19 pandemic, we performed sensitivity analyses using three analytical scenarios (Supplemental Table 9). When restricted to the pre-pandemic period (1999–2019), age-adjusted CVD mortality demonstrated a significant downward trend (APC − 2.07%, P = 0.0019), indicating sustained improvement before pandemic disruptions. In contrast, inclusion of the full study period (1999–2023) yielded a slight, non-significant increase (APC + 0.75%, P = 0.38). Notably, after excluding the pandemic years 2020–2021 while retaining all other years (1999–2019 and 2022–2023), the trend attenuated toward null (APC − 0.25%, P = 0.74). These findings suggest that the apparent recent rise in mortality was largely driven by deaths occurring during the pandemic rather than by a sustained pre-existing upward trajectory.

In summary, CVD mortality among MM patients exhibited overall stability over 25 years but demonstrated substantial underlying complexity, with divergent patterns across temporal phases, cardiovascular disease subtypes, and demographic and geographic strata (Fig. 8).

Fig. 8.

Overview of cardiovascular disease mortality patterns in multiple myeloma: A comprehensive 25-year analysis of trends, disparities, and clinical implications. Note: AAMR, age-adjusted mortality rate; APC, annual percent change; AAPC, average annual percent change; NH, non-Hispanic; * indicates statistically significant trend (P < 0.05). Rates per 100,000; age-adjusted to the 2000 U.S. standard population

Discussion

In this 25-year, population-level analysis of U.S. adults with MM, we documented three key findings that challenge the apparent stability suggested by the overall AAPC (+ 0.04%). First, joinpoint analysis revealed five temporally distinct phases masking dynamic fluctuations: an early rise (1999–2001), sustained decline (2001–2009), prolonged plateau (2009–2018), marked pandemic-era increase (2018–2021), and partial recent decline (2021–2023). Second, the 2018–2021 surge was characterized by substantial increases in treatment-related CVD subtypes—hypertensive disease (26.4% of total CVD deaths; AAPC + 4.79%), cardiac arrhythmias (14.4%; AAPC + 3.71%), heart failure (22.0%; AAPC + 0.80%), and thromboembolic events (4.6%; AAPC + 1.79%)—while traditional atherosclerotic endpoints declined (IHDs: 25.1%, AAPC − 1.15%; cerebrovascular disease: 9.8%, AAPC − 1.07%). Third, pronounced disparities persisted, with disproportionate burden among older adults, males, and NH Black individuals (8.38 vs. 3.51 per 100,000 for NH White adults). Given the ecological nature of death certificate data lacking individual-level treatment exposures and COVID-19 infection status, these temporal associations represent hypothesis-generating observations rather than established causal relationships.

Our national, long-term, joinpoint-based analysis extends prior single-center or trial-based reports by documenting temporal inflection points at the population level and identifying which CVD subtypes contributed most to the dynamic trends. Two recent population-based studies examined cardiovascular mortality in MM patients: Yin et al. [8] analyzed U.S. data through 2018 without joinpoint methodology or comprehensive CVD subtype stratification, while Eisfeld et al. [7] examined German mortality trends but did not extend into the pandemic era. Although prior clinical series and safety analyses, including SEER-Medicare analyses [9, 29], have documented clinically important cardiovascular adverse events (CVAEs) with modern MM therapies, particularly PIs and IMiDs, those studies focused on event incidence rather than mortality, examined specific treatment cohorts rather than population-wide patterns, or lacked our study’s 25-year span, six-category CVD subtype analysis, granular demographic stratification, and multi-level geographic resolution. The observed patterns—declining atherosclerotic endpoints alongside rising hypertensive, arrhythmic, thrombotic, and heart-failure mortality—reveal shifting cardiovascular risks as MM care and survivorship have evolved.

Our findings align with emerging international evidence. SEER-Medicare analyses document significantly elevated risks among MM patients versus controls for heart failure, arrhythmias, and venous thrombosis [9, 29]. European Myeloma Network consensus statements and pooled trial analyses report comparable burdens, with 60–70% of MM patients presenting cardiovascular comorbidities at diagnosis [10, 30]. However, notable regional variations in thromboembolic outcomes exist, potentially reflecting differences in thromboprophylaxis adherence and healthcare organization [15, 31]. These international comparisons demonstrate that cardiovascular outcomes in MM are modifiable through system-level interventions, underscoring the need for globally coordinated surveillance and prevention strategies.

Mechanistic interpretations

Our findings likely reflect five non-mutually exclusive mechanisms rather than direct causal effects of any single factor:

Therapeutic evolution and cardiotoxicity

The early-2000s CVD mortality decline temporally coincided with novel agent introduction—particularly PIs (bortezomib, FDA-approved 2003) and IMiDs—which substantially improved myeloma-specific survival. However, this temporal concordance warrants cautious interpretation. While these agents likely reduced overall myeloma mortality and thereby altered competing risks, later-generation PIs such as carfilzomib (approved 2012) carry significant cardiotoxicity that may offset cardiovascular benefits.

Carfilzomib’s expanding utilization temporally coincided with rising hypertensive disease, arrhythmias, and heart-failure deaths, though this association cannot establish causation. Cardiac arrhythmia mortality began increasing markedly after 2013, while hypertensive disease mortality surged during 2018–2021 as carfilzomib use expanded in both frontline and relapsed/refractory settings throughout the mid-to-late 2010s. Systematic reviews and pooled safety analyses report substantial rates of CVAEs with carfilzomib—with any-grade events in approximately one in five patients and grade ≥ 3 events in a nontrivial minority—often including hypertension, heart failure, and arrhythmias [13, 32]. Mechanistic studies demonstrate cardiomyocyte stress, endothelial dysfunction, and mitochondrial injury from proteasome inhibition [14, 33, 34], supporting potential treatment-related contributions to observed mortality patterns. However, temporal trends may also reflect evolving clinical practices. Heightened awareness of PIs cardiotoxicity may have improved cardiovascular event detection and reporting over time. Enhanced surveillance protocols and increased cardio-oncology collaboration could have increased ascertainment of cardiovascular causes on death certificates. Thus, observed increases in certain CVD mortality subtypes may partly reflect improved detection and documentation practices rather than solely true incidence increases, though death certificate data cannot disentangle these contributions.

The observed increase in arrhythmic deaths likely reflects multiple non-mutually exclusive pathways. First, drug-induced hypertension may promote left ventricular hypertrophy and diastolic dysfunction, creating electrophysiologic substrate for atrial fibrillation and ventricular arrhythmias. Second, PIs-related heart failure independently increases arrhythmic risk through neurohormonal activation, myocardial remodeling, and electrolyte shifts. Third, direct cardiotoxicity from mitochondrial dysfunction and altered calcium handling may exert pro-arrhythmic effects, potentially compounded by electrolyte disturbances (hypokalemia, hypomagnesemia) common in MM patients receiving corticosteroids and supportive therapies.

Thrombotic risk

IMiDs (thalidomide, lenalidomide, pomalidomide), particularly when combined with high-dose corticosteroids or multi-agent regimens, increase venous—and in some contexts arterial—thrombotic risk [15]. International practice guidelines recommend risk-stratified thromboprophylaxis [35]. However, the steady upward trend in thromboembolic mortality despite guideline availability suggests suboptimal prophylaxis implementation, including underutilization in community settings and inconsistent adherence. This gap between evidence-based recommendations and real-world practice underscores the ongoing need for improved prophylaxis uptake and systematic adherence monitoring.

Disease-intrinsic factors

MM biology itself contributes to cardiovascular risk: monoclonal proteins and light chains can cause cardiac amyloidosis and direct myocardial injury; systemic inflammation and hyperviscosity, electrolyte disturbances (e.g., hypercalcemia), and immobilization related to skeletal disease increase thrombotic and arrhythmic risks [36, 37]. Cardiac light-chain amyloidosis (AL), affecting approximately 10–15% of MM patients, remains an important, sometimes under-recognized driver of heart failure and sudden cardiac death in plasma-cell disorders [38].

Pandemic-era effects

The late-2010s/early-2020s spike temporally overlaps both increased uptake of potent agents and the COVID-19 pandemic, during which disruptions to routine care, delayed presentations, and infection-related cardiovascular complications were widely reported [39–41]. Sensitivity analyses provided quantitative evidence that the observed rise in CVD mortality was primarily driven by deaths occurring during the pandemic years (2020–2021), rather than by a sustained reversal in long-term trends. Pre-pandemic trends demonstrated significant mortality decline, which became non-significant when pandemic years were included. Critically, after excluding only the pandemic years while retaining post-pandemic data, the trend attenuated toward null, indicating rapid normalization after 2021. This pattern suggests a transient perturbation linked to pandemic-related factors—such as direct COVID-19 cardiovascular sequelae, healthcare system disruptions, and delayed cardiovascular care—rather than a fundamental shift in underlying mortality trajectories. Although the temporal coincidence between the broader use of second-generation PIs and increased cardiovascular deaths is biologically plausible, these findings should be interpreted as correlational rather than causal, given the overlapping pandemic influences. Establishing causality would require patient-level data linking treatment exposures, comorbidities, and infection status.

Competing risks

Improved myeloma survival created a larger, aging survivor pool at extended cardiovascular risk, potentially producing apparent CVD mortality increases independent of true risk changes. Observed patterns likely reflect interplay among treatment-related cardiotoxicity, compositional shifts as improved cancer control enables CVD manifestation, and changes in survivor age and comorbidity profiles.

Demographic and geographic disparities

The widening sex disparity—characterized by modestly increasing male mortality (AAPC + 0.18%) alongside declining female mortality (AAPC − 0.40%)—likely reflects multiple non-mutually exclusive mechanisms. First, males may receive more aggressive cardiotoxic regimens or demonstrate lower adherence to cardiovascular risk mitigation strategies. Second, sex-specific biological factors, including hormonal influences on cardiovascular susceptibility, pharmacokinetic differences, and sex-related variations in MM biology, may confer differential cardiotoxic vulnerability. Third, males historically exhibit lower engagement with preventive cardiovascular care and screening services. These divergent trajectories underscore the need for sex-tailored cardiovascular risk stratification and potentially intensified surveillance protocols for male MM patients receiving cardiotoxic therapies.

The steep age gradient in CVD mortality—rising progressively from younger (45–54 years) to oldest (≥ 85 years) adults—reflects the interplay of biological aging, treatment tolerance, and cumulative cardiovascular risk. Younger adults (45–54 years) demonstrated sustained mortality decline, likely attributable to superior physiologic reserve, lower baseline comorbidity burden, and more intensive cardiovascular surveillance that enabled safer tolerance of cardiotoxic regimens. Middle-aged groups (55–74 years) exhibited biphasic trajectories—initial decline through the early 2010 s followed by mid-to-late 2010 s increases and subsequent recent decline—temporally coinciding with the sequential therapeutic evolution from first- to second-generation PIs with heightened cardiotoxic potential. This age stratum represents a particularly vulnerable population: sufficiently young to receive intensive regimens yet old enough to harbor accumulated cardiovascular risk factors. Most concerning, adults ≥ 85 years experienced sustained mortality increases from 2015 onward, driven by cumulative cardiotoxic exposure in an aging survivor pool, high baseline cardiovascular burden, competing-risk dynamics whereby improved myeloma control unmasked cardiovascular mortality, and potential underutilization of cardio-oncology surveillance among frail elderly patients. These divergent age-specific trajectories underscore the imperative for age-tailored cardiovascular risk stratification, individualized treatment selection that balances oncologic efficacy against cardiotoxic liability, and enhanced supportive care for vulnerable older populations.

NH Black adults experienced a persistent 2.4-fold higher CVD mortality than NH White adults (8.38 vs. 3.51 per 100,000) throughout the 25-year period. Although temporal trends suggested potential slight convergence—with non-significant decline among Black patients (AAPC − 0.35%, 95% CI: −0.94% to + 0.52%) versus non-significant increase among White patients (AAPC + 0.16%, 95% CI: −0.11% to + 0.45%)—the absolute gap remained substantial without meaningful narrowing. This enduring inequity underscores deeply entrenched structural and systemic barriers, including differential access to specialized cardio-oncology care, socioeconomic disadvantage, insurance coverage gaps, and lower clinical trial participation rates [42–44]. Notably, these disparities likely reflect healthcare system failures rather than biological differences, consistent with national surveillance documenting persistent racial gaps in both cardiovascular outcomes and cancer care access [44]. Implementation research examining telemedicine approaches, patient navigation programs, and community outreach strategies may help identify effective interventions [45, 46], though such efforts require sustainable funding and healthcare system support.

Geographic patterns revealed complex regional variation. The regional disparities observed in our MM cohort may partially reflect broader U.S. cardiovascular mortality patterns. Recent national surveillance data document persistently elevated heart disease and stroke mortality across Southern states—the “Heart and Stroke Belt”—with comparatively lower rates in the Midwest and certain Northeastern regions [47], suggesting that population-level cardiovascular risk profiles, healthcare infrastructure, and social determinants contribute substantially to regional variation independent of myeloma-specific treatment factors.

Urbanization-specific mortality rates varied modestly (large metropolitan: 4.05; rural: 3.93; medium/small metropolitan: 3.54 per 100,000), yet all strata exhibited parallel temporal patterns of pre-2010s decline followed by late-2010s increases. The marginally elevated metropolitan mortality likely reflects tertiary center concentration treating higher-risk patients receiving intensive cardiotoxic regimens, while rural populations—despite subspecialty access barriers—may receive less aggressive treatment. These countervailing forces may explain the narrow urbanization disparities relative to pronounced racial/ethnic inequities.

State-level heterogeneity was substantial. Several states demonstrated significant late-2010s increases (Illinois, South Carolina, Oklahoma, Maryland, Minnesota, Wisconsin, Virginia, Mississippi), while others sustained long-term declines (New York, Connecticut, Missouri, West Virginia, Alabama, Ohio). This geographic mosaic likely reflects state-level variation in treatment paradigms, cardio-oncology infrastructure, thromboprophylaxis adherence, Medicaid expansion, and baseline cardiovascular burden. States maintaining sustained declines may have earlier implemented integrated cardio-oncology models or superior thromboprophylaxis protocols, warranting comparative health systems research to identify transferable cardiovascular risk mitigation strategies.

Clinical and public health implications

Our findings have potential clinical and public health implications, though their translation into practice should be considered within the context of observational study limitations and local resource availability.

First, these results support prioritizing comprehensive cardiovascular risk stratification before initiating therapies with established cardiovascular risks. For patients being considered for agents such as carfilzomib or IMiD-based combinations, structured baseline cardiovascular assessment—including detailed cardiac history, blood-pressure measurement, 12-lead electrocardiography, and consideration of echocardiography and cardiac biomarkers (e.g., high-sensitivity troponin, natriuretic peptides) when indicated—may facilitate shared decision-making regarding therapy choice and surveillance intensity [17, 18, 48, 49]. Validated risk stratification tools, including the HFA-ICOS cardio-oncology risk score and established cardiovascular risk calculators (e.g., SCORE2, Framingham), can help operationalize this assessment in clinical practice [50, 51]. Contemporary cardio-oncology guidelines recommend such risk-based assessment and surveillance to balance oncologic benefit and cardiovascular safety [52, 53], though implementation feasibility varies across healthcare settings.

Second, the observed increases in hypertensive events, arrhythmias, and heart failure mortality—particularly concentrated in early treatment cycles and among high-risk populations—suggest potential value in active monitoring strategies. These findings support consideration of protocol-driven blood pressure monitoring, periodic electrocardiography and electrolyte surveillance, and maintaining a low threshold for cardiology consultation or initiation of evidence-based cardioprotective therapies (e.g., ACE inhibitors/ARBs for blood pressure and cardiomyopathy prevention, beta-blockers for arrhythmias and heart failure) [17, 49] when abnormalities or symptoms emerge. Establishing structured cardio-oncology surveillance programs with serial cardiac biomarker assessment and periodic echocardiography may help detect and manage cardiovascular toxicity early, though resource requirements and cost-effectiveness of such programs warrant further evaluation. Multidisciplinary cardio-oncology pathways integrating hematology, cardiology, nursing, and pharmacy can help preserve oncologic treatment intensity while managing cardiovascular toxicity [54, 55].

Third, the persistent rise in thromboembolic deaths observed in this analysis supports continued emphasis on International Myeloma Working Group (IMWG)-endorsed VTE risk stratification and prophylaxis strategies in patients receiving IMiDs and steroid-containing regimens. Findings suggest that individualizing prophylaxis by patient risk (aspirin, low-molecular-weight heparin, or direct oral anticoagulants as appropriate) and reassessing across treatment phases may represent important practice considerations [15, 56]. While emerging evidence suggests DOACs may offer comparable efficacy to LMWH with advantages in convenience, head-to-head data in MM patients receiving IMiDs remain limited, and agent selection should be individualized based on bleeding risk, renal function, and drug interactions. Improving prophylaxis uptake and tailoring anticoagulation agents to individualized bleeding and thrombosis risk profiles remain areas warranting continued clinical attention [57].

Collectively, these population-level findings—documenting subtype-specific temporal increases, persistent demographic disparities, and pandemic-era vulnerabilities—provide epidemiologic evidence to inform future cardio-oncology guideline updates, particularly regarding risk stratification approaches, surveillance intensity recommendations for specific CVD subtypes, and survivorship care frameworks that prioritize cardiovascular health equity in MM populations.

Limitations

Several limitations constrain causal inference from our findings. First, death-certificate data are subject to misclassification and lack granular clinical information [58, 59]. Race and ethnicity designations may be misclassified, particularly for multiracial individuals and certain ethnic subgroups (e.g., Asian, Pacific Islander, Indigenous populations), potentially underestimating the true magnitude of disparities. Moreover, aggregation of diverse populations into the NH Other category (American Indian/Alaska Native and Asian/Pacific Islander) may mask important within-group heterogeneity in cardiovascular outcomes, though sample size limitations precluded further disaggregation. Cardiovascular comorbidities are frequently underreported on death certificates, which may lead to underestimation of overall CVD burden and misattribution of proximate causes of death. Our analysis included CVD as either underlying or contributing cause of death, which captures a broader spectrum of cardiovascular involvement than analyses restricted to underlying cause alone but introduces heterogeneity in the directness of CVD’s role in each death. We cannot ascertain individual-level drug exposures, dosing regimens, treatment sequences, comorbidities, or COVID-19 infection status, precluding causal inference and limiting our ability to disentangle treatment effects from pandemic-related influences.

Second, competing risks from declining MM-specific mortality substantially influence CVD mortality trend interpretation. Improved myeloma survival created a larger, aging survivor pool at extended cardiovascular risk, potentially producing apparent CVD mortality increases independent of true risk changes. Observed patterns likely reflect interplay among treatment-related cardiotoxicity, compositional shifts as improved cancer control enables CVD manifestation, and changes in survivor age and comorbidity profiles. This competing risk dynamic means that temporal increases in CVD mortality may paradoxically reflect successful cancer treatment enabling patients to survive long enough to experience cardiovascular events, rather than solely increased cardiovascular toxicity. Disentangling these mechanisms requires individual-level data with longitudinal follow-up from MM diagnosis, which our cross-sectional mortality surveillance data cannot provide.

Third, the CDC WONDER database does not permit adjustment for race/ethnicity, socioeconomic status, or other confounding factors that may contribute to geographic disparities beyond age standardization. Additionally, simultaneous stratification by race/ethnicity and geographic region was precluded by small cell sizes and CDC WONDER confidentiality restrictions (suppression of cells < 10 deaths), limiting our ability to identify specific geographic clustering of racial disparities that could inform place-based targeted interventions. Whether observed geographic differences persist after accounting for these demographic and social determinants remains uncertain and represents an important priority for future studies using linked cancer registry, claims, and electronic health record data that can support multivariable adjustment.

Fourth, state-level estimates in small-population jurisdictions were unstable and sometimes suppressed for confidentiality reasons. We did not apply spatial or temporal smoothing methods to preserve observed data structure and prevent artificial precision that might mask true local variations. Consequently, state-level findings should be interpreted cautiously, with particular attention to confidence interval widths provided in Supplemental Table 8.

Fifth, the multiple subgroup and subtype analyses conducted increase the likelihood of false-positive findings due to multiple statistical testing. We did not apply formal multiple-testing corrections (e.g., Bonferroni adjustment) to balance the risk of type I error against type II error in this descriptive surveillance study aimed at detecting clinically important trends. Consequently, isolated statistically significant findings in specific subgroups should be interpreted cautiously and validated in independent cohorts. We prioritized effect size magnitudes, confidence interval widths, and consistency of patterns across related subgroups to contextualize clinical significance beyond statistical significance thresholds. As an ecological study, our analysis cannot establish individual-level causal relationships between specific treatments and outcomes, nor account for unmeasured confounders such as baseline cardiovascular risk or treatment adherence.

Collectively, these limitations—including race misclassification, comorbidity underreporting, competing risk dynamics, and ecological inference constraints— underscore that observed temporal associations should be interpreted as hypothesis-generating correlations rather than causal attributions. However, these population-level data provide critical epidemiologic signals regarding temporal inflection points, subtype-specific mortality patterns, and demographic disparities that are difficult to detect in smaller clinical cohorts, thereby justifying and informing subsequent analytic studies using linked datasets with individual-level exposure and outcome data. Importantly, the large scale and national representativeness of this 25-year dataset ensure that these limitations do not invalidate the primary descriptive findings regarding temporal trends and demographic disparities, which remain essential for surveillance, hypothesis generation, and public health priority setting.

Future directions

To advance from association to attribution, future studies should link national death records with cancer registries and claims data (e.g., SEER-Medicare, electronic health record cohorts) to quantify individual-level exposure-outcome relationships with competing risk adjustment [60–62]. Based on our findings, such studies should prioritize hypertensive disease, cardiac arrhythmias, heart failure, and thromboembolic events—outcomes showing marked increases (AAPCs + 0.80% to + 4.79%) and collectively accounting for 63% of CVD deaths—as primary endpoints, with particular focus on cardiotoxic therapy exposures. Such linkage would enable multivariable adjustment for demographic and socioeconomic factors, clarifying whether regional disparities reflect patient populations, treatment access, or healthcare system characteristics. Prospective cardio-oncology cohorts and randomized or pragmatic trials evaluating monitoring and prevention strategies represent additional research priorities. Pragmatic designs embedded in routine care may be particularly suited for testing cardio-oncology interventions in heterogeneous, real-world MM populations with comorbidities. Such studies might examine blood pressure management algorithms, biomarker-guided surveillance protocols, or prophylactic anticoagulation algorithms in high-risk IMiD recipients [63–65]. Mechanistic research on PIs cardiotoxicity and pharmacogenomic modifiers of cardiovascular risk will help identify highest-risk patients.

Conclusions

In this 25-year national analysis, overall cardiovascular mortality among patients with MM remained stable; however, sharp increases in hypertensive, arrhythmic, thromboembolic, and heart failure deaths during 2018–2021 coincided with expanded cardiotoxic therapy use and pandemic-related disruptions, disproportionately affecting older and NH Black adults. These findings underscore the need for systematic cardiovascular risk assessment before high-risk therapy, adherence to guideline-based thromboprophylaxis, and integration of cardio-oncology care into routine MM management.

Targeted interventions are also needed to address persistent racial and geographic disparities through equitable access to multidisciplinary cardiovascular support. Beyond clinical practice, these results may inform development of national quality metrics and survivorship programs prioritizing cardiovascular surveillance and risk mitigation in MM. Sustained collaboration among oncologists, cardiologists, and public health systems will be essential to preserve survivorship gains while minimizing cardiovascular harm in the modern treatment era.

Abbreviations

AAPC

Average annual percent change

AAMR

Age-adjusted mortality rate

AL

Light-chain amyloidosis

APC

Annual percent change

CDC

Centers for Disease Control and Prevention

CMR

Crude mortality rate

CVAE

Cardiovascular adverse event

CVD

Cardiovascular disease

ICD-10

International Classification of Diseases, Tenth Revision

IHD

Ischemic heart disease

IMiDs

Immunomodulatory drugs

IMWG

International Myeloma Working Group

MM

Multiple myeloma

NH

Non-Hispanic

PIs

Proteasome inhibitors

U.S.

United States

VTE

Venous thromboembolism

wBIC

Weighted Bayesian Information Criterion

WONDER

Wide-Ranging Online Data for Epidemiologic Research

Authors’ contributions

Ying Tian: Conceptualization; methodology; data curation; formal analysis; visualization; writing – original draft; writing – review and editing. Xiaobin Guo: Conceptualization; supervision; methodology; validation; project administration; resources; writing – review and editing. Yue Zhang: Conceptualization; supervision; validation; project administration; writing – review and editing; writing – original draft.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability

The datasets generated and/or analyzed during the current study are available on the CDC Wonder Database, [https://wonder.cdc.gov/](https://wonder.cdc.gov).

Declarations

Ethics approval and consent to participate

This study used publicly available, de-identified mortality data from CDC WONDER and was exempt from institutional review board approval and informed consent requirements.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.