Two Decades of Stroke in the United States: A Healthcare Economic Perspective

Stroke is a leading cause of morbidity and the fifth highest cause of mortality in the USA. It impacts the financial status of patients, families, and healthcare systems [1]. Globally, it is the second-leading cause of death and the third-leading cause of lost disability-adjusted life years [2]. Strokes are broadly categorized as either ischemic or hemorrhagic. Management depends on the severity and subtype of stroke, and its subspecialized care can be expensive [1]. Patients may undergo fibrinolytic therapy, thrombectomy, craniectomy, repeated imaging studies and often are monitored for complications in an intensive care unit level of care. Complications such as blood pressure extremes, brain tissue edema, brain rebleeding or new-onset bleeding, vasospasm, seizures, and severe disabling symptoms can affect short- and long-term outcomes from a physical, emotional, and economical perspective. In recent years, there has been a significant effort in healthcare to provide an efficient treatment while considering the financial aspects of the treatment plans. Stroke is regarded as a medical condition with significant morbidity and mortality. Understanding the evolving financial impact this diagnosis has is crucial for future planning. This study aims to assess the economic implications of stroke on the healthcare systems for the past 2 decades and project the impact for the next decade.

This cross-sectional study of inpatient subjects from 2000 to 2020 with stroke was collected from the Healthcare Cost and Utilization Project (HCUP) across 50 states. HCUP is the largest healthcare database and includes all states in the USA, which includes a stratified sample of discharges from community hospitals in the USA [3]. We queried patients admitted primarily for ischemic or hemorrhagic stroke, including first-time and recurrent stroke patients. HCUP has ICD-9 data available from 2000–September 2015 and ICD-10 data from October 2015 to 2020. We utilized HCUP’s Clinical Classification Software (CCS) category “109: acute cerebrovascular disease” and the HCUP CCS Refined categories “CIR020: cerebral infarction” and “CIR021: acute hemorrhagic cerebrovascular disease.” Patients were evaluated for demographics, length of stay (LOS), mortality, and hospital charges and costs. Statistical Z-testing with a significance of p < 0.05 was conducted for the analysis, and exponential and linear trends were performed for the forecasts. Analyses were repeated for subgroups: sex, age group, expected payer, patient community-level income, and race and ethnicity. The University of Texas Health Science Center at San Antonio exempted the project from the full review due to the anonymous nature of the HCUP database.

During the study period, 12,158,747 stroke subjects were studied, with 51.9% female and a mean age of 70.08 (±0.16) years old. The mean rate of stroke discharges per 100,000 persons was 187.71 (±3.44), decreasing from 200 to 193 during the study (p = 0.16), as shown in Figure 1. The mean percentage of deaths was 8.78% (±0.17), which decreased from 10.96% to 6.81% (p = 0.00), as shown in Figure 2. The mean LOS was 6.28 days (±0.08), which increased from 6.70 to 7.15 (p = 0.00).

During the study period, the aggregated national bill was USD 725 billion, while the aggregated cost to hospitals was USD 188 billion. The mean hospital charges per patient were USD 57,178 (±1,504), increasing from USD 19,647 to USD 121,765 per person during the study period (p = 0.00), as shown in Figure 3. These data closely conform to an exponential growth pattern and forecasting per patient charges for the next 10 years demonstrates a cost of USD 287,836 by 2030. Table 1 portrays the annual bill for stroke throughout the study period with forecasts to 2030, which is projected to be USD 208 billion. Concomitantly, the mean hospital costs per stay were USD 15,781 (±330), which increased from USD 8,946 to USD 27,225 (p = 0.00).

Mean results by subgroups are shown in Table 2. Females had a higher prevalence of stroke hospitalization and mortality (p < 0.05), while there was no significant difference between sexes regarding LOS, hospital charges, and costs. Patients aged 65–84 had the highest prevalence of stroke and mortality, while patients aged 1–17 had highest hospital charges, costs, and LOS. Expected payer for 65% of stroke patients was Medicare: in addition to highest prevalence, this subgroup had the highest number of inhospital deaths. Medicaid patients had the highest hospital charges, costs, and LOS. After “Missing” and “Other” expected payer categories, “Self-pay/no charge” patients had the highest mean mortality rate at 9.94 (±0.62). Patient community-level incomes were available from 2003 to 2020, which utilizes a patient’s ZIP code of residence as a proxy measure of socioeconomic status. Low income is defined as household income less than two-thirds the national median income (2020: < USD 52,000), middle income as two-thirds to double it (2020: USD 52,000–USD 156,000), and high income as greater than double it (2020: > USD 156,000) [3, 4]. In our study, middle-income patients had the highest mean annual number of discharges and in-hospital deaths. Low-income patients had the highest rate of discharges per 100,000 persons. After the “Missing” category, high-income patients had the highest mean hospital charges and costs, while low-income patients had the longest LOS. Patient race/ethnicity was available from 2018 to 2020. White, non-Hispanic patients had the highest number of stroke patients and inhospital deaths. Black, non-Hispanic patients had the highest rate of discharges per 100,000 persons and longest LOS. Hispanic patients had the highest mean hospital charges per stay, yet Asian/Pacific Islander, non-Hispanic patients had the highest costs to the hospital per stay.

Before delving into discussing the results, it is important to understand our terminology. “Charges” refer to what the hospital bill insurance, while “costs” reflect the actual cost of providing care. In general, cost is one-third of the charges, although price transparency is desperately needed in US healthcare. Insurance companies and healthcare systems negotiate deals to reconcile these bills and make “payments” that are typically in the midrange between cost and charge. However, payers vary widely in their payment amounts; for example, Medicaid and Medicare both pay below cost. To simplify, costs can be seen as the financial burden on the hospital, while charges are the burden on the patient and their insurance, while payments are what the hospital receives from the payer [5].

Figure 1 depicts stroke frequency in the past 2 decades, with notable dips in 2007 and 2020. A potential explanation to the apparent 2007 trend reversal is the advent of the stroke center. The American Heart Association (AHA) started to define stroke centers in 2000, and US hospitals subsequently began to organize primary stroke centers, which were first certified by the Joint Commission in December 2003. By 2009, approximately 600 primary stroke centers were functional in the USA, which correlates with our observed rise of the stroke population during the same time [6, 7]. The decline in 2020 is likely secondary to the COVID-19 pandemic and its associated restrictions.

Our data show that the number of stroke patients was steadily climbing in the 2010s until reaching an uncharacteristic low in 2020. The first documented COVID-19 case in the USA was in January 2020, and the ensuing lockdown restrictions and fear of virus exposure may have discouraged patients to present to the hospital with findings concerning for stroke. It has also been suggested that some stroke patients deferred emergency services and died at home during the pandemic, and even others may have presumed that their hospitals and clinics were closed to non-COVID-19 patients [8]. A multinational study found that stroke services were affected by the pandemic, with younger patients, higher NIHSS at admission and mRS at discharge, higher hemorrhagic stroke prevalence, longer door-to-needle time, decreased tPA administration, and increased thrombectomy frequency [9]. A US study found that patients were more likely to present outside of eligible treatment windows in 2020, but it did not find significant changes in stroke outcomes [10]. The lockdowns and simplification of routines may have encouraged medication adherence and control of chronic disease states, as well as potentially decreased access to high sodium meals prepared in dining establishments. Travel and outing restrictions yielded less opportunity for exertion injuries, falls, and motor vehicle accidents, as well as decreased transmission of respiratory pathogens known to exacerbate preexisting conditions [8]. However, it may also be true that fewer outpatient visits were scheduled/attended; patients may not have received their routine preventive care or stroke follow-up.

Figure 2 shows a decline in stroke mortality, juxtaposed with a concurrent rise in associated charges and costs seen in Figure 3. The mortality improvement is multifactorial. Many advances have occurred in the last 20 years, including the establishment of more comprehensive stroke centers and the evolution of treatment approaches like thrombectomy and thrombolysis. A Belgian study found mechanical thrombectomy to be a highly cost-effective treatment, providing quality-adjusted survival while saving EUR 56.2 million over 5 years (USD 61.3 million), while a UK study found that GBP 17,221 (USD 21,763) were saved per thrombectomy patient compared to matched controls [11, 12]. Recent studies suggest that endovascular therapy alone is more cost-effective than combined with TPA in acute ischemic stroke [13]. Paradoxically, charges per patient have surged by over sixfold, reaching a national total nearly on par with the annual Medicare budget [14]. This increase is possibly multifaceted, including inflation, more medical workups, and higher medical costs in recent years. HCUP does not correct for inflation in its data, but online supplementary Table A (for all online suppl. material, see https://doi.org/10.1159/000536011) outlines the changes in Annual Consumer Price Index (CPI) and annual inflation rate from 2000 to 2020, including the cumulative inflation rate during the study period. The CPI, which is a measure of the average change over time in the prices paid by urban consumers for a market basket of consumer goods and services, increased from USD 172.20 in 2000 to USD 258.81 in 2020, translating to a cumulative inflation rate of 50.3% during the same period [15].

HCUP does not divulge details regarding Medicaid, Medicaid, and private insurance coverage of stroke patients, nor does it offer insight about specific treatment differences based on expected payer. These features can be assumed to vary greatly among the more than 1,000 US insurers [5]. However, a separate HCUP study found that private insurance was associated with improved clinical outcomes, lower mortality, and shorter LOS. Private insurance had fewer patient safety indicators when compared to patients with Medicare, Medicaid, or self-pay [16]. Further discourse is needed concerning price transparency and perceived care gaps among patients with diverse payers.

The significance of preventive medicine by primary care internists, such as controlling hyperlipidemia, hypertension, diabetes, and smoking cessation, cannot be underestimated. This conclusion aligns with previous studies’ calls to action, including a policy statement from the AHA and American Stroke Association in 2013 [17]. Their paper aimed to project future costs of stroke, and their direct annual costs increased from USD 71.55 billion in 2012 to USD 184.13 billion in 2030. Although this AHA paper includes additional stroke-related care costs, our data follow a similarly dramatic trend: USD 31.84 billion in 2012 to USD 208.25 billion in 2030.

Other costs contributing to stroke’s disease burden include non-healthcare and indirect costs, including transportation, relocation, informal care, and lost productivity. Rochmah et al. [18] posit that in addition to the disease burden, there is a massive national economic loss secondary to stroke treatment and prolonged recovery period. The 2023 article by Gerstl et al. [19] focuses on macroeconomic losses, calling attention to the massive value of lost welfare and stroke’s major public health implications. Recent global cost-effective successes include Australia’s Victorian Stroke Telemedicine program and Norway’s Mobile Stroke Unit; however, these are reactive rather than preventive strategies, ever emphasizing the need for preventive policies [20, 21]. The 2022 article by Owolabi et al. [22] calls attention to the growing burden on low- and middle-income countries and outlines specific implementation strategies for primary stroke prevention.

Though outpatient costs were excluded from this analysis, it is essential to emphasize the significance of quantifying and researching poststroke care in the context of discussing the financial implication posthospital discharge. Long-term disability and complications from stroke play a crucial role in the financial burden to both healthcare systems and patients. In addition, conducting additional studies that assess the management of risk factors in ambulatory settings can offer a more comprehensive cost-benefit analysis of preventive measures. In summary, while recent years have witnessed improvement in stroke outcomes, LOS, and lower mortality, the escalating economic impact of stroke demands proactive measures from healthcare policymakers.

The authors thank the Healthcare Cost and Utilization Project (HCUP) for its availability of compiled healthcare data.

Ethics approval was not required for this study as it involved a public database and no human or animal subjects.

The authors have no conflicts of interest to declare.

This research has received no funding.

Miss Lorio is the corresponding author for this work and was involved in all abstract and manuscript drafts. She acquired and interpreted the initial data from HCUP to create an original work. Dr. Garcia-Rodriguez contributed significant edits to all abstract and manuscript drafts and helped in the initial formulation of ideas. Dr. Seifi was crucial to the original design of the work and assisted in all abstract and manuscript drafts, as well as data acquisition and interpretation.

All datasets are publicly and freely available at HCUPnet [3].