Cost of Cardiovascular Disease Event and Possible Cardiovascular Disease Treatment-Related Complication Hospitalizations in the United States

Abstract

Background:

Cardiovascular disease (CVD) is among the costliest conditions in the US and cost-effectiveness analyses can be used to assess economic impact and prioritize CVD treatments. We aimed to develop standardized, nationally representative CVD event and selected possible CVD treatment-related complication hospitalization costs for use in cost-effectiveness analyses.

Methods:

Nationally representative costs were derived using publicly available inpatient hospital discharge data from the 2012–2018 National Inpatient Sample. Events were identified using principal ICD-9 and 10 codes. Facility charges were converted to costs using charge-to-cost ratios and total costs were estimated by applying a published professional fee ratio. All costs are reported in 2021 US dollars. Mean costs were estimated for events overall and stratified by age, sex, and survival status at discharge. Annual costs to the US healthcare system were estimated by multiplying the mean annual number of events by the mean total cost per discharge.

Results:

The annual mean number of hospital discharges among CVD events was highest for heart failure (1,087,000 per year) and cerebrovascular disease (800,600 per year). The mean cost per hospital discharge was highest for peripheral vascular disease ($33,700 [95% CI: $33,300 – $34,000]) and ventricular tachycardia/ventricular fibrillation ($32,500 [95% CI: $32,100 – $32,900]). Hospitalizations contributing the most to annual US healthcare costs were heart failure ($19,500 million; 95% CI $19,300 – $19,800 million) and acute myocardial infarction ($18,300 million; 95% CI: $18,200 – $18,500 million). Acute kidney injury was the most frequent possible treatment complication (515,000 per year) and bradycardia had the highest mean hospitalization costs ($17,400 [95% CI: $17,200 – $17,500]).

Conclusions:

The hospitalization cost estimates and statistical code reported in the current study have the potential to increase transparency and comparability of cost-effectiveness analyses for CVD in the US.

Introduction

Cardiovascular disease (CVD) is the leading cause of death in the United States (US) and among the costliest health conditions, accounting for $320 billion in direct health care costs in 2016.^{1, 2} While age-standardized CVD death rates decreased steadily from the mid 1900s to early 2000s, they began to rise from 2010–2020.^3–5 Even if recent favorable age-standardized CVD incidence and mortality were to be sustained, aging of the Baby Boomer generation is expected to result in increased absolute prevalence of CVD, which will increase the volume of CVD hospitalizations, and ultimately further increase the total cost burden on the US healthcare system.⁶ These factors highlight the urgent need to economically manage CVD in US adults.

Cost-effectiveness analyses can be used to quantify the expected health and economic impacts of CVD treatments and prioritize the highest-value interventions.^{7, 8} In the US, however, there is no organization that standardizes or regulates cost-effectiveness analyses, which may lead to substantial variability in the methods and input parameters used, including costs. Standardizing the source and approach used to derive the inputs used in cost-effectiveness analyses, in particular the cost of CVD event hospitalizations and hospitalizations for possible complications of CVD treatment, will increase the comparability of results across studies.

The purpose of this study is to report nationally representative CVD event and selected possible CVD treatment-related complication hospitalization costs using a standardized approach in the National Inpatient Sample (NIS) and estimate the contribution to annual healthcare costs in the US. We aimed to estimate the mean costs of these hospitalizations in the US overall and by subgroups of sex, age, and survival status at discharge from 2012–2018. We also explore ways NIS analyses can be tailored to provide parameter estimates suited to specific needs in cost-effectiveness analyses, such as stratifying total hospitalization costs by procedures performed during a hospitalization.

Methods

Data Source

We used the publicly available NIS database of inpatient hospital discharges from the Agency for Healthcare Research and Quality (AHRQ) from 2012–2018.⁹ NIS includes data on over seven million visits to community hospitals per year. When properly weighted, NIS provides nationally representative utilization and charge data for over 35 million discharges from US hospitals per year.⁹ Charges in the NIS include data from all payer types (e.g., Medicare, Medicaid, private insurance, self-pay) and can be converted to facility costs using charge-to-cost ratios (CCRs) for each hospital in the sample. Each hospital discharge in the NIS includes but is not limited to information on patient demographics, International Classification of Diseases (ICD) diagnosis and procedure codes, Healthcare Common Procedure Coding System (HCPCS) and Current Procedural Terminology (CPT) codes, total facility charges, discharge status, and length of stay. However, the NIS does not include rehabilitation or long-term acute care hospital discharges. The data for this study are publicly available and do not meet the definition of human participants research. As such, institutional review board approval was not indicated.

Hospital Event Identification

The following CVD event types were included in this analysis: acute myocardial infarction (AMI), other coronary heart disease (CHD), heart failure (HF), cerebrovascular disease, sudden cardiac arrest, atrial fibrillation, peripheral vascular disease (PVD), and ventricular tachycardia/ventricular fibrillation (VT/VF). Selected possible CVD treatment-related complications included hypotension, syncope, bradycardia, electrolyte abnormality, acute kidney injury (AKI), and major bleeds.^{10, 11} Table S1 presents the ICD-9 and ICD-10 codes used to categorize the CVD event and possible CVD treatment-related complication types. We identified CVD and possible CVD treatment-related complication discharges in NIS using the principal ICD-9 diagnosis code from first quarter 2012 to third quarter 2015 and ICD-10 diagnosis codes from fourth quarter of 2015 to fourth quarter of 2018. The mean costs for each event in each quarter of 2015 and 2016 were examined to assess the potential impact of changing from ICD-9 to ICD-10 and subsequent coding practices during 2015. To highlight ways in which NIS analyses can be customized, we also explored the impact of stratifying hospitalizations for AMI by revascularization procedure performed if any. AMI events were stratified into the following categories: AMI with no percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG), AMI with PCI, and AMI with CABG. For this exploratory analysis, the procedure code may have been in any position.

Total Hospitalization Costs

We converted the facility charges reported in NIS to costs using the CCRs provided by AHRQ with the NIS data. The CCRs are hospital specific and are based on all-payer inpatient costs. We then inflated costs to 2021 US dollars before analysis using the healthcare component of the Personal Consumption Expenditure Index available from the US Bureau of Economic Analysis.¹² Total hospitalization costs were estimated by multiplying the facility costs in 2021 US dollars by a published professional fee ratio.^{13, 14} The professional fee ratio was developed for use with charge data from publicly available discharge datasets, such as NIS, and was derived using 2012 MarketScan insurance claims data.¹⁴ We used the overall commercial insurance adjusted professional fee ratio of 1.264 for discharges.¹⁴ Facility costs can be calculated by dividing total costs by the professional fee ratio.

Statistical Analysis

All analyses were performed using Stata version 16 (StataCorp LP, College Station, TX) and figures generated using R version 4.0 (Vienna, Austria). To facilitate adoption of a standardized approach to estimating CVD event and possible CVD treatment-related complication hospitalization costs, the Stata code used for this analysis is included in the Supplemental Material. All analyses accounted for the complex survey weighting design of NIS, and the estimates presented are representative of the national averages of US adults aged ≥18 years of age. To provide cost-effectiveness analysis with nationally representative “off-the-shelf” estimates for CVD event and possible CVD treatment-related complication hospitalizations, we present overall means and sex-, survival status-, and age- (18–34, 35–44, 45–54, 55–64, 65–74, 75–84, and ≥85 years) stratified estimates. To ensure that stable estimates were achieved, we ensured that each age, sex, and survival stratified group had at least 50 unweighted discharges. When there were <50 unweighted discharges, we combined adjacent age groups. We calculated the average annual US healthcare costs of each CVD event from 2012 to 2018 by multiplying the estimated annual number of events in the US by the mean total cost per hospital discharge. Additionally, we calculated the mean cost of each CVD hospitalization type per year and US healthcare costs for each CVD event per year. We also used generalized linear regression models with a gamma distribution and a log link function to estimate the total cost of hospital discharges for each event type given selected covariates. A standardized model was developed in one event type (i.e., AMI) and then applied to each of the other CVD and possible CVD treatment-related complications. The standardized model included the following terms: categorical age (18–34, 35–44, 45–54, 65–74, 75–84, and ≥85 years), sex, survival status at discharge, and an interaction term between categorical age and survival status to develop a broadly applicable, parsimonious model. Model fit was assessed by comparing the ratio of the predicted mean to the observed mean total cost overall, and stratified by sex, survival status, and age. To demonstrate how the regression models could be modified to adjust for other covariates or improve model fit, additional covariates (i.e., race/ethnicity, number of chronic conditions, and length of stay) were sequentially added to the standardized model for AMI.

Results

The number of CVD events, mean hospitalization costs, and annual US healthcare costs, by event type, overall and by sex, captured by the NIS between 2012–2018 are presented in Table 1. When weighted, the estimated annual number of CVD hospitalizations varied from 11,971 for sudden cardiac arrest to 1,086,914 for HF. PVD hospitalizations were the most expensive CVD event (mean $33,700 [95% CI: $33,300 – $34,000]). Several other CVD event types had a mean cost greater than $25,000 per hospitalization including AMI ($29,500 [95% CI: $29,300 – $29,700]), other CHD ($29,900 [95% CI: $29,700 – $30,200]), and VT/VF ($32,500 [95% CI: $32,100 – $32,900]). Atrial fibrillation was the least expensive event on average ($13,100 [95% CI: $13,100 – $13,200]). The cost of hospitalizations was similar for male and female patients for most CVD event types. However, HF, AMI, and other CHD hospitalizations were 16%, 18%, and 21% more expensive, respectively, for males compared with females.

Table 1.

Cardiovascular disease hospitalization costs, overall and by sex in the United States from 2012–2018.

	Total Hospitalizations, Unweighted N	Mean Total Cost per Hospitalization (95% CI)*	Average No. of Hospitalizations Per Year†	Annual US Healthcare Costs (in millions; 95% CI)
Heart failure
Overall	1,521,700	$18,000 ($17,800 – $18,200)	1,086,914	$19,500 ($19,300 – $19,800)
Female	734,500	$16,600 ($16,400 – $16,700)	524,614	$8,700 ($8,600 – $8,800)
Male	787,200	$19,300 ($19,000 – $19,500)	562,300	$10,800 ($10,700 – $11,000)
Cerebrovascular disease
Overall	1,120,900	$22,300 ($22,000 – $22,500)	800,614	$17,800 ($17,600 – $18,000)
Female	559,500	$22,100 ($21,900 – $22,300)	399,629	$8,800 ($8,700 – $8,900)
Male	561,400	$22,400 ($22,200 – $22,700)	401,000	$9,000 ($8,900 – $9,100)
Acute myocardial infarction
Overall	869,400	$29,500 ($29,300 – $29,700)	620,986	$18,300 ($18,200 – $18,500)
Female	332,100	$26,600 ($26,400 – $26,800)	237,214	$6,300 ($6,300 – $6,400)
Male	537,300	$31,300 ($31,100 – $31,500)	383,771	$12,000 ($11,900 – $12,100)
Atrial Fibrillation
Overall	652,100	$13,100 ($13,100 – $13,200)	465,814	$6,100 ($6,100 – $6,200)
Female	323,900	$12,400 ($12,300 – $12,500)	231,371	$2,900 ($2,900 – $2,900)
Male	328,200	$13,800 ($13,700 – $14,000)	234,443	$3,200 ($3,200 – $3,300)
Other coronary heart disease
Overall	579,400	$29,900 ($29,700 – $30,200)	413,886	$12,400 ($12,300 – $12,500)
Female	197,600	$26,300 ($26,100 – $26,500)	141,129	$3,700 ($3,700 – $3,700)
Male	381,900	$31,800 ($31,500 – $32,000)	272,757	$8,700 ($8,600 – $8,700)
Peripheral vascular disease
Overall	286,600	$33,700 ($33,300 – $34,000)	204,743	$6,900 ($6,800 – $7,000)
Female	107,300	$32,400 ($32,000 – $32,800)	76,629	$2,500 ($2,500 – $2,500)
Male	179,300	$34,400 ($34,100 – $34,800)	128,100	$4,400 ($4,400 – $4,500)
Ventricular tachycardia or fibrillation
Overall	84,900	$32,500 ($32,100 – $32,900)	60,643	$2,000 ($1,900 – $2,000)
Female	23,100	$31,100 ($30,600 – $31,700)	16,486	$500 ($500 – $500)
Male	61,800	$33,000 ($32,500 – $33,500)	44,157	$1,500 ($1,400 – $1,500)
Sudden Cardiac Arrest
Overall	16,800	$24,500 ($23,900 – $25,000)	11,971	$300 ($300 – $300)
Female	7,500	$23,200 ($22,400 – $24,000)	5,386	$100 ($100 – $100)
Male	9,200	$25,500 ($24,700 – $26,200)	6,586	$200 ($200 – $200)

CI=confidence interval.

^*Total hospitalization costs were estimated by applying a published professional fee ratio to the facility costs estimated from the National Inpatient Sample.

^†Mean number of hospitalizations per year in the US when incorporating the survey weighting in the National Inpatient Sample.

Among CVD events, the impact on annual US healthcare costs were determined to be the highest for HF ($19,500 million; 95% CI $19,300 – $19,800 million), AMI ($18,300 million; 95% CI: $18,200 – $18,500 million), and cerebrovascular disease ($17,800 million; 95% CI: $17,600 – $18,000 million). Cost of CVD events gradually rose over time from 2012 to 2018 aside from PVD and sudden cardiac arrest which increased more rapidly from 2014 to 2018 (Figure 1). The total cost to the US healthcare system for HF, AMI, cerebrovascular disease, and PVD increased over time while CHD costs decreased (Figure 2). For each quarter (i.e., Q1, Q2, Q3, and Q4), costs of AMI, CHD, HF, and cerebrovascular disease were similar for years 2015 and 2016 (Table S2).

Figure 1.

Mean cardiovascular disease event hospitalization costs over time.

Abbreviations: AMI – acute myocardial infarction; Cerebrovascular – cerebrovascular disease; CHD – coronary heart disease; HF – heart failure; SCA – sudden cardiac arrest; AFib – Atrial Fibrillation; PVD – peripheral vascular disease; VT/VF – ventricular tachycardia/ventricular fibrillation

Figure 2.

Total United States healthcare costs from cardiovascular disease hospitalizations over time.

Abbreviations: AMI – acute myocardial infarction; Cerebrovascular – cerebrovascular disease; CHD – coronary heart disease; HF – heart failure; SCA – sudden cardiac arrest; AFib – Atrial Fibrillation; PVD – peripheral vascular disease; VT/VF – ventricular tachycardia/ventricular fibrillation

Tables S3-6 present costs of AMI events stratified by sex, survival status, and age, and by revascularization procedure performed (i.e., overall, no PCI or CABG, with PCI, with CABG). The mean cost of an AMI event with no PCI or CABG was the least expensive at $21,898. Having an AMI with PCI cost $31,522 and an AMI with CABG ($71,788) was more than triple the cost of an AMI with no PCI or CABG.

Tables S7-S13 present costs of individual CVD event types (other CHD, HF, cerebrovascular disease, sudden cardiac arrest, atrial fibrillation, PVD, and VT/VF) stratified by sex, survival status, and age. CVD event hospitalization costs among patients who died were higher for all conditions aside from sudden cardiac arrest (Tables S3-S13). For sudden cardiac arrest, costs among adults who survived were $44,447 and costs among adults who died were $18,712 (Table S10). For other CHD, HF, atrial fibrillation, and PVD, hospitalization costs for patients who died were more than twice as much as patients who did not die (Tables S7, S8, S11, S12). Other CHD hospitalizations were particularly costly among adults who died compared to those who did not: $79,047 and $29,534, respectively (Table S7). Costs of CVD events exhibited a parabolic increase and decrease as patient age increased.

Table 2 presents the number and cost of hospitalizations for possible CVD treatment-related complications, overall and by sex. Of the possible CVD treatment-related complications examined, AKI was the most frequent with an estimated 515,071 weighted annual events, and hypotension was the least frequent (weighted n=97,271). Bradycardia hospitalizations were the most expensive possible CVD treatment-related complication ($17,400 [95% CI: $17,200 – $17,500]). Electrolyte imbalance and syncope were the least expensive possible CVD treatment-related complications, costing $9,600 (95% CI: $9,500 – $9,700) and $9,600 (95% CI: $9,500 – $9,600), respectively. The costs of possible CVD treatment-related complications were similar for male and female patients for all complication types. Tables S14-S19 present costs of individual possible CVD treatment-related complication types stratified by sex, survival status, and age.

Table 2.

Selected possible CVD treatment-related complication hospitalization costs, overall and by sex in the United States from 2012–2018.

	Total Hospitalizations, Unweighted N	Mean Total Cost per Hospitalization (95% CI)*	Average No. of Hospitalizations Per Year†
Acute kidney injury
Overall	721,100	$13,200 ($13,200 – $13,300)	515,071
Female	352,600	$13,000 ($12,900 – $13,100)	251,886
Male	368,400	$13,400 ($13,400 – $13,500)	263,171
Major Bleed Event
Overall	550,000	$14,300 ($14,200 – $14,300)	392,829
Female	260,700	$13,800 ($13,700 – $13,900)	186,186
Male	289,300	$14,700 ($14,600 – $14,800)	206,643
Electrolyte Abnormality
Overall	509,200	$9,600 ($9,500 – $9,700)	363,686
Female	293,400	$9,400 ($9,300 – $9,500)	209,557
Male	215,800	$9,900 ($9,800 – $10,000)	154,129
Syncope
Overall	184,500	$9,600 ($9,500 – $9,600)	131,786
Female	98,500	$9,400 ($9,400 – $9,500)	70,343
Male	86,000	$9,800 ($9,700 – $9,800)	61,457
Bradycardia
Overall	160,300	$17,400 ($17,200 – $17,500)	114,486
Female	86,500	$17,000 ($16,900 – $17,100)	61,757
Male	73,800	$17,800 ($17,600 – $17,900)	52,729
Hypotension
Overall	136,200	$10,500 ($10,400 – $10,600)	97,271
Female	65,400	$10,400 ($10,300 – $10,500)	46,729
Male	70,800	$10,600 ($10,500 – $10,700)	50,543

CI=confidence interval.

^*Total hospitalization costs were estimated by applying a published professional fee ratio to the facility costs estimated from the National Inpatient Sample.

^†Mean number of hospitalizations per year in the US when incorporating the survey weighting in the National Inpatient Sample.

Tables S20 and S21 present β coefficients for regression equations for costs of CVD events and possible CVD treatment-related complications, respectively. Table S22 presents β coefficients for sequentially constructed models with additional covariates (i.e., race/ethnicity, number of chronic conditions, and length of stay) to the standardized model for AMI. Tables S23 and S24 present the ratio of mean predicted costs based on regression equations to mean observed costs for CVD events and possible CVD treatment-related complications. The majority of predicted costs were within 1% of observed costs.

Discussion

Using nationally representative data from the NIS, the current study estimated costs of CVD event and selected possible CVD treatment-related complication hospital discharges and the associated annual US healthcare costs from 2012 to 2018. We estimated that, while HF was not the most expensive CVD event per hospitalization, it was the most commonly occurring and so had the largest overall impact on US healthcare costs. Other conditions like AMI, cerebrovascular disease, and other CHD were both common and expensive per individual hospitalization, leading them to contribute substantially to US healthcare costs. We also provide cost estimates for possible CVD treatment-related complication hospitalizations (hypotension, syncope, bradycardia, electrolyte abnormality, major bleeds, and acute kidney injury). These cost estimates, the associated cost calculation methods, regression equations to predict costs, and statistical code can be used in future cost-effectiveness studies and inform healthcare policy decisions.

Similar to previous literature, the current study estimated HF to be a leading cause of hospital admissions and result in substantial healthcare spending.^{15, 16} With the prevalence of HF projected to increase over the next decade, economic management is increasingly important.^{17, 18} The expense and frequency of HF admissions and readmissions (defined as return to hospital for HF within 30 days following discharge after HF treatment) has led to insurer financial incentives to reduce readmissions and innovative care models focused on avoiding preventable admissions and readmissions. Proven interventions to lower HF hospital admission rates include comprehensive transitional care planning before hospital discharge,¹⁹ and a team-based care approach to outpatient HF management including nurses, pharmacists, or other non-physicians to implement guideline-directed medical treatments.^{20, 21} Integrating these best practices into quality standards, quality improvement programs, and reimbursement criteria have the potential to curb the health and economic impacts of HF over time.

AMI hospitalizations were one of the most expensive CVD events. The NIS data allowed us to stratify AMI cost analysis by procedure type: high mean AMI cost was driven by the high cost of PCI and CABG procedures, while AMI hospitalizations without procedures were relatively inexpensive. We found that mean cost of AMI per hospitalization did not rise appreciably over the period 2012 to 2018. A recent analysis of Medicare data over a similar period (2000 to 2018) found that overall spending on and utilization per beneficiary of inpatient and outpatient revascularization procedures decreased over this interval (40.7% decrease in CABG utilization per Medicare enrollee; 26.4% decrease for PCI).²² The authors of the study hypothesize that increased use of guideline recommended medical management may have resulted in lower spending on PCI and CABG procedures during this timeframe. PVD hospitalizations were the most expensive event ($33,700 [95% CI: $33,300 – $34,000]), which may be due to vascular surgery and procedures with these hospitalizations and warrants further research. There was an increase in PVD and sudden cardiac arrest costs from 2014 to 2018. This could be due to many reasons (e.g., changes in coding practices during and after the transition from ICD-9 to ICD-10 codes) and additional research is needed to further examine the underlying factors.

Cost-effectiveness analyses provides a value framework for policy and healthcare decision makers to assess if the health gains with new interventions are worth the economic costs.^{7, 8} The current study contributes to the growing body of literature attempting to provide transparent, reproducible, and consistent model inputs when estimating the cost-effectiveness of CVD treatments.^{23, 24} Morey et al. calculated total annual health expenditures using publicly available Medical Expenditure Panel Survey (MEPS) data for CVD that overlaps with our analysis. NIS may be considered a preferable source for inpatient hospitalizations over MEPS due to the number of hospitalizations included. For example, Morey et al. used MEPS data from 2011 to 2016 and reported an unweighted N of 905 individuals with HF. Using NIS from 2012 to 2018, we identified over 1.5 million HF hospital discharges hospitalizations, which allows for greater confidence when examining subgroups. Additionally, we focused on acute care hospitalization costs, while Morey et al. present a complementary composite cost that includes hospital inpatient stays, emergency room visits, outpatient visits, office visits, and home health costs. Researchers can use the “off-the-shelf” estimates provided in the current study for their own research, and, similar to Morey et al., we include our statistical code in the supplemental material to increase transparency and reproducibility of our approach. In addition to calculating event costs, we also provide examples in our code of how to calculate costs for subcategories within events (i.e., cost of AMI with or without CABG or PCI). This can be applied to other event types such as PVD where surgical interventions may or may not be performed. We also provide regression models to predict hospitalization costs, which could be used “off-the-shelf” or tailored to researchers’ needs. Though the regression models generally predicted mean costs within 1% of observed costs, researchers should consider the sample size of cells when using the regression model approach as the range for the ratio of predicted to actual costs increased in some cells likely due to small sample sizes. If these approaches to derive model inputs or a common set of model cost inputs are consistently adopted by cost-effectiveness analyses of CVD treatments, even if only in scenario analysis, it will increase the comparability of analyses and provide healthcare decision and policy makers confidence that differences in cost-effectiveness estimates are not due to differences in model inputs.

A strength of our approach is the use of the NIS which is a publicly available and nationally representative dataset. Using the NIS data also allowed us to calculate national costs across payers. The current study has limitations. Though the NIS includes hospital wide charge-to-cost ratios, it is important to note that there may be variation in charge-to-cost ratios within hospitals by cost-center (e.g., these hospital wide ratios may overestimate costs for surgical patients).²⁵ Other publicly available data sources could be used to generate national hospitalization cost estimates, such as the mean cost per discharge paid by the Centers for Medicare and Medicaid Services among individuals with fee-for-service Medicare coverage. Hospital discharges can be identified by Medicare Severity Diagnosis Related Group (MS-DRG) and the data include average total payments, which includes the copayment and deductible amounts paid by patients among other components. These estimates may be insufficient for research needs in some cases as the MS-DRGs allow limited stratification (e.g., stratify by with or without major complication or comorbidity but not by age or sex). Further, these estimates do not include other payer types (e.g., private insurers) and largely excludes younger and middle-aged adults as Medicare primarily covers those aged ≥65 years of age. Additionally, it is estimated that Medicare pays hospitals about half what private payers do. NIS captures hospital discharges but does not include acute or chronic rehabilitation costs, outpatient office visits and testing, and pharmaceuticals, whether prescription or over the counter. This may lead to an underestimation of event-related costs if not otherwise accounted for in cost-effectiveness analyses. Another limitation is that many of the selected possible CVD treatment-related complications included in our analysis could also occur as part of the natural history of disease independent of treatment or be associated with other treatments. For instance, in addition to CVD treatments like the use of beta blockers, bradycardia can also be caused by hypothyroidism, myocarditis, and sleep apnea. Because this analysis did not attempt to disaggregate attribution to CVD treatment vs other causes for the complications, we did not estimate annual US healthcare costs for these events. Sharp increases in the cost per PVD and sudden cardiac arrest hospitalization observed during 2012–2018 may be related to changes in ICD code definitions related to these conditions, leading to billing practice changes, rather than a sudden increase in procedural costs or changes in clinical practice. Careful examination of the number of hospital discharges and mean costs per quarter in 2015 through 2016 is warranted when researchers include NIS data spanning the ICD code transition. The NIS data also has several limitations related to this study.^{26, 27} NIS does not include US state in the sampling framework; thus, state-specific estimates or cross-state comparisons require another dataset. Also, individual patients cannot be identified in the NIS.⁹ If the cost or frequency of rehospitalizations are needed, another dataset, such as the National Readmissions Database, must be used. Also, NIS does not account for indirect costs, including home care giver expenses, transportation, and lost labor productivity.

Conclusions

In the 21^st century, epidemiologic and demographic forces are combining to increase the absolute prevalence of CVD in the US population. The resulting increase in volume of associated hospitalizations will ultimately multiply the total cost burden on the US healthcare system.⁶ As spending on CVD hospitalizations is responsible for 15% of total US healthcare spending,² efforts to spend more wisely by prioritizing use of effective and high-value treatments, disincentivizing low-value care, making hospitalization more efficient, and taking measures to prevent early re-hospitalization are needed. The hospitalization cost estimates, the associated cost calculation methods, and statistical code reported here have the potential to increase transparency and comparability of the cost-effectiveness analyses needed for the US to progress toward strategic cost-containment and highest-value care for patients.

Supplementary Material

Supplemental Publication Material

Supplemental Methods: Stata code

Table S1. Diagnosis and procedure codes for cardiovascular disease and treatment related severe adverse events.

Table S2. Mean hospitalization costs for select cardiovascular disease events in each quarter of 2015 and 2016.

Table S3. Unadjusted mean survey-weighted mean hospitalization cost of acute myocardial infarction hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S4. Unadjusted mean survey-weighted mean hospitalization cost of acute myocardial infarction hospitalizations with no percutaneous coronary intervention or coronary artery bypass graft in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S5. Unadjusted mean survey-weighted mean hospitalization cost of acute myocardial infarction hospitalizations with percutaneous coronary intervention in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S6. Unadjusted mean survey-weighted mean hospitalization cost of acute myocardial infarction hospitalizations with coronary artery bypass graft in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S7. Unadjusted mean survey-weighted mean hospitalization cost of other coronary heart disease/angina hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S8. Unadjusted mean survey-weighted mean cost of heart failure hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S9. Unadjusted mean survey-weighted mean cost of cerebrovascular disease hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S10. Unadjusted mean survey-weighted mean cost of sudden cardiac arrest hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S11. Unadjusted mean survey-weighted mean cost of atrial fibrillation hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S12. Unadjusted mean survey-weighted mean cost of peripheral vascular disease hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S13. Unadjusted mean survey-weighted mean cost of ventricular tachycardia or ventricular fibrillation hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S14. Unadjusted mean survey-weighted mean cost of hypotension hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S15. Unadjusted mean survey-weighted mean cost of syncope hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S16. Unadjusted mean survey-weighted mean cost of bradycardia hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S17. Unadjusted mean survey-weighted mean cost of electrolyte abnormalities hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S18. Unadjusted mean survey-weighted mean cost of acute kidney injury hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S19. Unadjusted mean survey-weighted mean cost of major bleed hospitalizations in the United States stratified by age, survival status, and sex from 2012–2018 (2021 USD).

Table S20. Generalized linear model functions to predict the cost of hospital discharges for cardiovascular disease events.

Table S21. Generalized linear regression model functions to predict the cost of hospital discharges for selected possible CVD treatment-related complications.

Table S22. Generalized linear regression model functions to predict the cost of hospital discharges for acute myocardial infarction including additional covariates.

Table S23. Ratio of mean predicted to mean observed costs for hospital discharges for cardiovascular disease events.

Table S24. Ratio of mean predicted to mean observed costs for hospital discharges for selected possible CVD treatment-related complications.

What is known:

• Cardiovascular disease (CVD) is among the costliest conditions in the United States, and the cost is expected to rise over the next several decades.

• Cost-effectiveness analyses can be used to prioritize the highest-value CVD interventions, but there is no standardized approach to calculating the cost of CVD events used in these analyses.

What the study adds:

• Using a publicly available, nationally representative dataset, we provide updated costs of CVD hospitalizations and hospitalizations for possible complications of CVD treatment.

• These cost estimates, the associated cost calculation methods, and statistical code used to calculate these costs can be used in cost-effectiveness studies and inform healthcare policy decisions.

Non-standard Abbreviations and Acronyms:

NIS

National Inpatient Sample

AHRQ

Agency for Healthcare Research and Quality

CCRs

charge-to-cost ratios

HCPCS

Healthcare Common Procedure Coding System

CPT

Current Procedural Terminology

PVD

peripheral vascular disease

VT/VF

ventricular tachycardia/ventricular fibrillation

AKI

acute kidney injury

PCI

percutaneous coronary intervention

CABG

coronary artery bypass graft

MEPS

Medical Expenditure Panel Survey

MS-DRG

Medicare Severity Diagnosis Related Group