Is the Affordable Care Act Medicaid Expansion Linked to Change in Rate of Ventricular Assist Device Implantation for Blacks and Whites?

Abstract

Background

The Affordable Care Act (ACA) has been associated with increased heart transplant listings among Blacks, who are disproportionately uninsured. It is unclear whether the ACA is also associated with increased ventricular assist device (VAD) implantation in Blacks.

Methods

Using Healthcare Cost and Utilization Project Data State Inpatient Databases (HCUP SID) from 19 states and Washington D.C., we analyzed 1,157 patients from early adopter states (ACA Medicaid Expansion by January 2014) and 785 patients from non-adopter states (no implementation from 2013 to 2014). Piecewise Poisson regression with a discontinuity was used to estimate change in census-adjusted rates of VAD implants by race and ACA adopter status one year before and after January 2014.

Results

Following the ACA Medicaid Expansion, the proportional change in rate increased significantly among Blacks from early adopter [1.40(95%CI:1.12,1.75), pre 0.57/100,000 to post ACA 0.80/100,000] but not non-adopter states [1.25(95%CI:0.98,1.58), pre 0.40/100,000 to post ACA 0.50/100,000]. However, the early and non-adopter changes in implantation rates were not statistically different from each other (p=0.50). There were no immediate changes in Whites in either state group following the ACA Medicaid Expansion [early adopter 1.12(95%CI: 0.98,1.29), pre 0.27/100,000 to post ACA 0.30/100,000; non-adopter 0.98(95%CI: 0.82,1.16), pre 0.27/100,000 to post ACA 0.26/100,000].

Conclusions

Among eligible states participating in HCUP SID, the ACA was not associated with immediate changes in VAD implantation rates by race. Although a significant increase in implantation rate was observed among Blacks from early adopter states, the change was not statistically different from the change seen in non-adopter states.

Blacks have the highest prevalence of heart failure compared to other racial groups¹ with two to three-fold higher risk of developing heart failure than Whites, particularly before the age of 65.¹ Similarly, heart failure mortality rates are highest among Blacks¹ and have increased significantly in the past decade among Black men.² Yet, racial and ethnic minorities receive less than a third of advanced heart failure therapies.³

Lack of insurance may contribute to inequitable allocation of advanced heart failure therapies. Before a patient can receive advanced heart failure therapy in the U.S., the patient must demonstrate ability to pay through either insurance or self-pay.⁴ This is a hurdle for racial minorities who are disproportionately uninsured.⁵

Increased access to insurance may reduce racial disparities in allocation of advanced heart failure therapies. Implementation of the U.S. Patient Protection and Affordable Care Act (ACA) reduced the uninsured population by nearly 40%,⁶ with racial and ethnic minorities experiencing the greatest proportional gains in insured status.⁵ The ACA Medicaid Expansion was associated with an increase in heart transplant listings among Blacks residing in states that adopted the policy but not other racial groups.⁷ Therefore, we hypothesized that the ACA Medicaid Expansion would be associated with increased ventricular assist device (VAD) implantation in Blacks.

METHODS

Data Source

Patients who received durable ventricular assist devices were identified through the Healthcare Cost and Utilization Project State Inpatient Databases (HCUP SID). HCUP SID is sponsored by the Agency for Healthcare Research and Quality and provides data to support health policy.⁸ HCUP SID includes unweighted inpatient discharge data from 97% of hospitals of participating states.⁸ Data include demographics such as self-reported race and ethnicity, sex, age group starting with 20 years of age, insurance type, procedure, and diagnostic codes. This study was approved by HCUP SID and each state. This study was deemed exempt from review by the University of Arizona Institutional Review Board due to deidentified data.

Study Population

Inclusion criteria included adults, 20 years of age and older, who had a hospitalization discharge with a ventricular assist device implantation code (containing International Classification of Diseases (ICD) 9 code 37.66, ICD 10 code 02HA0QZ, or current procedural terminology code 33979). Patients were excluded for missing race (n=38), income quartile (n=49), patient residence type [metro versus non-metro (defined by urban influence code of the U.S. Department of Agriculture’s Economic Research Service) n=6], or insurance (n=10). Discharge quarter and year were used for time of procedure since procedure date was not consistently available across states. Patients with race or ethnicity of Black (n=597) and Non-Hispanic White (n=1345) were included. Hispanics (n=119), Asians (n=41), American Indians (n=13), and other race (n=133) were excluded due to insufficient numbers.

We selected 19 states plus Washington D.C. and stratified states by ACA Medicaid Expansion adopter status (Figure 1). Early adopter (n= 12 states plus Washington D.C.) represents states which implemented ACA Medicaid Expansion by January 2014 and non-adopter (n=7 states) represents states with no implementation through end of study in December 2014. The remaining states were excluded for (Supplemental Methods): adoption of ACA Medicaid Expansion after implementation date of January 2014 or adoption with reduced coverage (n=7 states),⁹ missing data for race during any year of the study (n= 4 states), lack of participation or data availability through HCUP SID at time of data purchase spring 2018 (n=16 states), and prohibitive cost (n= 4 states). U.S. Census Bureau annual estimates were used to adjust VAD rates for racial demographics⁷ since populations vary by state (Supplemental Table 1).

Figure 1. State Groups Stratified by ACA Medicaid Expansion Adopter Status.

Blue states represent early adopters of the ACA Medicaid expansion; red states, non-adopters; grey, excluded.

Outcomes

The primary outcome was the change in census-adjusted rates of VAD implants immediately before and after ACA Medicaid Expansion implementation (January 2014). Secondary outcomes included proportional change in census-adjusted rates of VAD implants at time of ACA Medicaid Expansion according to insurance type and sex.

Statistical Analysis

Characteristics of VAD recipients were compared according to adopter status using chi-squared tests. Counts of VAD implants were aggregated by discharge quarter, race, and state group (early-adopter or non-adopter). Census estimates from July 1st of each year (2012 to 2014) were taken, summing estimates for ages 20+ years separately for early-adopter and non-adopter states and for Whites and Blacks. Quarterly estimates were obtained via linear interpolation, with July 1st assumed to be the mid-point of the year and the date for each quarter taken at the mid-point of the quarter. This was performed combining sexes and separately for each sex.

A piece-wise linear Poisson regression was conducted using the glm() function in R¹⁰ with a discontinuity at the time of ACA implementation (1 January 2014). Model terms included race (Black or White), state group (early-adopter or non-adopter) as well as the interactions of race and state group and interactions to allow the discontinuity effect to differ for each race and state group. Rates of VAD implantation were assumed to be constant within each year. To account for differences in population size, the natural log of the quarterly-interpolated census population estimate was included as an offset. A population size of 100,000 was used to obtain predicted rates of VAD implantation standardized per 100,000 persons. In addition to conducting the Poisson regression on the overall dataset, separate analyses were conducted on the following subsets: Medicare, Medicaid, private insurance, men, and women. For the sex-specific analyses, sex-specific population estimates were used in the offset. As a sensitivity analysis, the highest and lowest income quartiles were examined using similar analyses.

RESULTS

Population Characteristics

Among VAD recipients, 1,157 were from early adopter state groups and 785 from non-adopters (Table 1). Blacks represented a lower proportion of patients from early adopter states than non-adopter states (27.8%, 35.03%, p<0.01 respectively). The majority of patients from both groups were under the age of 65. In early adopter states, there was a higher proportion of Medicaid and private insurance beneficiaries and a lower proportion of Medicare beneficiaries. VAD recipients from early adopter states were more often from high-income quartiles than non-adopter states.

Table 1.

Black and White VAD Population from 2013–2014

Characteristic	Early Adopter N=1,157 %	Non-Adopter N=785 %	P-value
Age Group			0.57
20–34	7.17	6.50
35–49	19.62	19.87
50–64	41.66	39.36
65+	31.55	34.27
Sex			0.41
Men	77.53	75.92
Women	22.47	24.08
Race			<0.01
Black	27.83	35.03
White	72.17	64.97
Insurance			<0.01
Medicaid	12.45	8.79
Medicare	44.51	56.31
Other	2.94	3.57
Private	40.10	31.34
Income Quartile			<0.01
1 Lowest Income Quartile	17.29	39.24
2	20.05	32.61
3	26.36	19.11
4 Highest Income Quartile	36.30	9.04
Patient Residence Type			<0.01
Rural	11.58	17.45
Metro	88.42	82.55
Heart Failure			0.11
Systolic	82.97	85.86
Diastolic	1.47	1.78
Other	10.29	9.17
None	5.27	3.18
Anemia	49.44	57.71	<0.01
Atrial Arrhythmia	37.86	38.98	0.62
Chronic kidney disease	29.04	31.85	0.19
End Stage Renal Disease	2.68	2.68	>0.99
Coagulation defects	15.64	15.16	0.77
COPD	8.56	10.32	0.19
Depression	5.36	5.10	0.80
Diabetes	19.36	25.35	<0.01
Hypertension	40.02	41.91	0.40
Ischemic heart disease	39.50	45.86	0.01
Obese	7.43	9.30	0.14
Ventricular arrhythmia	45.72	44.59	0.62

COPD indicates chronic obstructive pulmonary disease. Ventricular arrhythmia includes ventricular tachycardia, paroxysmal ventricular tachycardia, ventricular fibrillation and flutter. Other heart failure represents diagnosis of heart failure (ICD9: 428.x, 402.01, 402.11, 402.91, 398.91, 404.01, 404.03, 404.11, 404.13, 404.91, 404.92 ICD10: I50.x, I132, I130, I110, I09.81) that was not also described as systolic or diastolic heart failure. None represents diagnosis of VAD but had no diagnosis of heart failure as defined by preceding sentence.

Outcomes

The per capita rate of VAD was higher in Blacks than in Whites (Table 2, Figure 2). At the time of ACA implementation, the rate of VAD implantation increased significantly by an estimated 40% among Blacks from early adopter states, from a pre ACA estimate of 0.57/100, 000 to post ACA 0.80/100,000 [proportional change in rate 1.40(95%CI: 1.12, 1.75)]. There was no statistically significant change in implant rates among Blacks from non-adopter states [proportional change in rate 1.25(95%CI: 0.98, 1.58), pre 0.40 to post ACA 0.50/100,000]. However, the proportional change in rates were not statistically different for Blacks from early adopter and non-adopter state groups (p=0.50). At the time of ACA implementation, VAD implant rates did not change significantly among White patients from either state group [early adopter 1.12 (95%CI: 0.98, 1.29), pre 0.27 to post ACA 0.30/100,000; non-adopter 0.98 (95%CI: 0.82, 1.16), pre 0.27 to post ACA 0.26/100,000] (Table 2, Figure 2). The proportional change in rates was not statistically different for Whites from early adopter and non-adopter state groups (p=0.22).

Table 2.

Impact of ACA Medicaid Expansion on VAD Rates

	Average Rate Per 100,000 Pre ACA (95% CI)		Instantaneous proportional change in rate January 1, 2014 ACA Implementation (95% CI)		Average Rate Per 100,000 Post ACA (95% CI)
	Early Adopter	Non-Adopter	Early Adopter	Non-Adopter	Early Adopter	Non-Adopter
Black	0.57 (0.48,0.68)	0.40 (0.33,0.48)	1.40* (1.12,1.75)	1.25 (0.98,1.58)	0.80 (0.69,0.92)	0.50 (0.43,0.59)
White	0.27 (0.24,0.30)	0.27 (0.24,0.30)	1.12 (0.98,1.29)	0.98 (0.82,1.16)	0.30 (0.27,0.33)	0.26 (0.23,0.29)
Medicaid
Black	0.11 (0.08,0.16)	0.06 (0.03,0.09)	1.51 (0.92,2.48)	1.39 (0.74,2.58)	0.17 (0.12,0.23)	0.08 (0.05,0.12)
White	0.02 (0.02,0.03)	0.01 (0.01,0.02)	1.43 (0.91,2.25)	1.15 (0.54,2.41)	0.03 (0.02,0.04)	0.02 (0.01,0.03)
Medicare
Black	0.22 (0.17,0.29)	0.20 (0.15,0.25)	1.51* (1.07,2.15)	1.38 (0.99,1.93)	0.34 (0.27,0.42)	0.27 (0.22,0.33)
White	0.12 (0.10,0.14)	0.16 (0.14,0.19)	1.17 (0.96,1.43)	0.88 (0.70,1.10)	0.14 (0.12,0.16)	0.15 (0.12,0.17)
Private
Black	0.23 (0.17,0.30)	0.14 (0.10,0.19)	1.23 (0.85,1.76)	0.89 (0.57,1.38)	0.28 (0.22,0.36)	0.12 (0.09,0.17)
White	0.12 (0.10,0.14)	0.08 (0.06,0.10)	0.99 (0.80,1.22)	1.15 (0.85,1.56)	0.12 (0.10,0.14)	0.09 (0.07,0.11)
Men
Black	0.83 (0.67,1.02)	0.56 (0.45,0.70)	1.54* (1.18,2.02)	1.22 (0.91,1.64)	1.27 (1.08,1.50)	0.69 (0.56,0.84)
White	0.44 (0.40,0.50)	0.46 (0.40,0.52)	1.11 (0.95,1.29)	0.96 (0.79,1.16)	0.49 (0.44,0.55)	0.44 (0.38,0.50)
Women
Black	0.35 (0.26,0.47)	0.26 (0.19,0.35)	1.10 (0.73,1.65)	1.30 (0.88,1.93)	0.39 (0.29,0.51)	0.34 (0.26,0.44)
White	0.10 (0.08,0.13)	0.09 (0.06,0.12)	1.17 (0.86,1.58)	1.06 (0.70,1.61)	0.12 (0.10,0.15)	0.09 (0.07,0.12)

^*denotes p<0.05

Figure 2. Estimated Rate of VAD Implantation by Race.

Dots represents model estimates of VAD implantation rates pre and post ACA Medicaid Expansion (blue lines indicate early adopters of the ACA Medicaid expansion, n=1157; red lines, non-adopters, n=785; vertical lines, 95% Confidence Bands).

When stratified by insurance group, only Black Medicare beneficiaries from early adopter states exhibited a significant increase in rate at the time of ACA implementation [early adopter 1.51 (95%CI: 1.07,2.15), pre 0.22 to post 0.34/100,000; non-adopter 1.38 (95%CI: 0.99,1.93), pre 0.20 to post 0.27/100,000; Table 2; Figures 3a, 3b, 3c]. However, the proportional change in rates in the early adopter and non-adopter states were not significantly different (p=0.71) for Black Medicare beneficiaries. When stratified by sex, only Black men from early adopter states exhibited a significant increase in VAD rates at the time of the ACA implementation [early adopter 1.54 (95%CI 1.18,2.02), pre 0.83 to post 1.27/100,000; non-adopter 1.22 (95% CI: 0.91,1.64), pre 0.56 to post 0.69/100,000; Table 2; Figures 4a, 4b], but proportional change in rates of early adopter and non-adopter were not significantly different from each other (p=0.25). When stratified by income quartile group (lowest and highest), there were no instantaneous differences in VAD implantation rates for each race group in 2014 (Supplemental Figures 1a, 1b).

Figure 3. Estimated Rate of VAD Implantation by Race Across Insurance Groups.

Patients are represented according to 3a. Medicaid insurance ( n=144 early adopter, n=69 non-adopter), 3b. Medicare insurance ( n=515 early adopter, n=442 non-adopter), and 3c. private insurance ( n=464 early adopter, n=246 non-adopter). Dots represents model estimates of VAD implantation rates pre and post ACA Medicaid Expansion (blue lines indicate early adopters of the ACA Medicaid expansion; red lines, non-adopters; vertical lines, 95% Confidence Bands).

Figure 4. Estimated Rate of VAD Implantation by Race Across Sex.

Patients are represented according to sex: 4a. men (n=897 early adopter, n=596 non-adopter), and 4b. women (n=260 early adopter, n=189 non-adopter). Dots represents model estimates of VAD implantation rates pre and post Medicaid Expansion(blue lines indicate early adopters of the ACA Medicaid expansion; red lines, non-adopters; vertical lines, 95% Confidence Bands).

DISCUSSION

In a study of 19 states and Washington D.C., the ACA Medicaid Expansion was not associated with immediate changes in VAD implantation by race. A significant instantaneous increase in VAD rates was observed among Blacks from early adopter states, but this change was not statistically different from the change observed in non-adopter states. Similarly, a significant increase in VAD rates was observed among Black Medicare beneficiaries and Black men from early adopter states, but changes were not significantly different from those in non-adopter states.

The per capita rate of VAD implantation was higher in Blacks than Whites. VAD rates are approximately two-fold higher in Blacks than Whites.³ Similarly, the prevalence of heart failure in Blacks is approximately double that of Whites.¹ However, Blacks with heart failure have approximately three- fold higher risk of death than Whites, particularly before the age of 65.² This disparity in rates of death may be related to a lower than expected receipt of advanced heart failure therapies.

This HCUP SID study is different than others that observed reduced racial disparities in association with the ACA Medicaid Expansion.⁵ A noteworthy difference may be related to a lack of participation of populous states in HCUP State Inpatient Database such as California and Texas. The lack of significant differences may also be related to ACA changes that affected both early adopter and non-adopter state groups. The ACA expanded access to insurance through Marketplace (public insurance supported by ACA which has no income exclusions), prohibited insurance companies from denying individuals with pre-existing conditions, and provided tax credits to help reduce insurance premiums and reduce patients’ out of pocket cost sharing.¹¹ Thus, during early implementation of the ACA, 20 million individuals gained access to insurance, of which the majority were White.⁶ However, Blacks and Hispanics had the greatest proportional increase in insurance due to higher proportions of uninsured minorities.^6,12 The ACA Medicaid Expansion was associated with increased likelihood that racial minorities would seek clinical care because of reduced cost.¹³ Heart transplant listings increased by 30% among Blacks residing in early adopter states and was significantly different from non-adopter states.⁷ Similarly, there was a significant increase in VAD rates among Blacks from early adopter states, but the increase was not significantly different from Blacks in non-adopter states. Adoption of the ACA Medicaid Expansion has the potential to reduce racial disparities since non-adopter states have higher proportions of uninsured individuals and Blacks than early adopter states.^6,14

Neither Black nor White Medicaid beneficiaries experienced an increase in VAD implants after the ACA Medicaid Expansion. This may be secondary to multiple factors. First, Medicaid beneficiaries across the U.S. have unequal coverage.¹⁵ Each state regulates what Medicaid will cover, and not all states cover VAD, particularly destination therapy implantation.^16,17 This differs from federally regulated Medicare, which covers both bridge to transplant and destination VAD.¹⁸ This may explain the rise in VAD rates among Black patients with Medicare. The HCUP SID database does not include VAD strategy, such as bridge to transplant VAD or destination VAD, but over 40% of VADs allocated to Blacks were destination therapy according to an Interagency Registry of Mechanically Assisted Circulatory Support study.³ The lack of universal Medicaid coverage of VAD destination therapy may contribute to continuing racial disparities in that indication. In addition, Medicaid beneficiaries are not universally eligible for subspecialty cardiovascular care,¹⁹ which may disparately affect racial minorities.

Second, the population of underinsured patients is growing, particularly while the ACA Medicaid Expansion is in flux.⁶ Most gains in access occurred during the first few years of the policy implementation,⁶ but some healthcare professionals refused Medicaid payors or limited new patients with Medicaid or Marketplace payors from the start of ACA expansion.^20,21 Since 2017, racial disparities have increased particularly in non-adopter states.⁶ Additional increases in uninsured status during this time period may be related to lack of awareness of affordable insurance opportunities, prohibitive costs for insurance,⁶ and legislation to reduce eligibility.²²

Last, advanced heart failure therapies allocation remains low in women irrespective of insurance access.⁷ Black women have nearly double the prevalence of heart failure than White women and men, higher prevalence than Black men, and higher prevalence of heart failure with reduced ejection fraction than White women.¹ National trends reveal that VAD rates have increased in Black women but remain below expected levels compared to other races and men.³ The lack of change in VAD rates among Black women in this HCUP SID study of health care policy suggests that additional factors contribute to disparities by race and sex. Recent literature has revealed racial bias in allocation of advanced heart therapies.²³ Men with identical clinical and social vignette presentations were allocated to advanced heart therapies according to race. The Black man was perceived to have greater issues with adherence and social support and thus was allocated to VAD, whereas the White man was perceived to have better adherence and was allocated to a heart transplant during discussions with healthcare professionals.²³ The intersection of racial and sex disparities is unclear and requires additional investigation.

Limitations

This study has several limitations. First, all states do not participate in HCUP SID, and the results may not be generalizable to the entire United States. Although patterns are similar to national trends of VAD implants by race,³ HCUP SID lack data for some populous states that could contribute to underpowered results. Second, we are unable to adjust for baseline comorbidities since the prevalence of each disease is not known within the census population for each state. Third, the role of patient and healthcare professional decision-making and cultural approaches to providing care are unknown for this study and could disparately affect allocation of VAD by race in the different state groups.

Conclusions

Among the HCUP SID, the ACA Medicaid Expansion was not associated with immediate changes in VAD implantation by race. Although a significant rise in VAD rates was observed among Blacks from early adopter states, this change was not statistically different from the change in Blacks from non-adopter states. There were no significant changes among Whites from any state group. Additional changes in healthcare policy may be indicated for reduction in racial disparities of VAD implantation.

What Is New

• A statistically significant increase in VAD implantations was observed only in Blacks from early adopter states, but it was not statistically different from the change observed in non-adopter states. Therefore, racial changes in VAD implantation rate cannot be attributed to the ACA timing.

• No immediate changes were observed in Whites from early adopter or non-adopter state groups.

Clinical Implications

• Insurance is a requisite for advanced heart failure therapies. U.S. racial groups are not equally insured. Additional advances in health care policy are needed to achieve racial parity in VAD implantation rates.

ABBREVIATIONS

ACA

Affordable Care Act

COPD

chronic obstructive pulmonary disease

HCUP SID

Healthcare Cost and Utilization Project Data State Inpatient Databases

ICD

International Classification of Diseases

VAD

ventricular assist device