Racial Disparities in Stroke Readmissions Reduced in Hospitals with Better Nurse Staffing

Abstract

Background:

Racial minorities are disproportionately affected by stroke, with Black patients experiencing worse post-stroke outcomes than White patients. A modifiable aspect of acute stroke care delivery not yet examined is whether disparities in stroke outcomes are related to hospital nurse staffing levels.

Objectives:

To determine whether 7- and 30-day readmission disparities between Black and White patients were associated with nurse staffing levels.

Methods:

We conducted a secondary analysis of 542 hospitals in four states. Risk-adjusted, logistic regression models were used to determine the association of nurse staffing with 7-day and 30-day all-cause readmissions for Black and White ischemic stroke patients.

Results:

Our sample included 98,150 ischemic stroke patients (87% White, 13% Black). Thirty-day readmission rates were 10.4% (12.7% for Black, 10.0% for White patients). In models accounting for hospital and patient characteristics, the odds of 30-day readmissions were higher for Black than White patients. A significant interaction was found between race and nurse staffing, with Black patients experiencing higher odds of 30-day and 7-day readmissions for each additional patient cared for by a nurse. In the best-staffed hospitals (< 3 patients per nurse), Black and White stroke patients’ disparities were no longer significant.

Discussion:

Disparities in readmissions between Black and White stroke patients may be linked to the level of nurse staffing in the hospitals where they receive care. Tailoring nurse staffing levels to meet the needs of Black ischemic stroke patients represents a promising intervention to address systemic inequities linked to readmission disparities among minority stroke patients.

Stroke is a leading cause of death and disability in the United States, with an estimated 795,000 new or recurrent strokes diagnosed annually (Tong et al., 2019). Racial minorities are disproportionately affected by stroke, with Black patients experiencing twice the incidence of stroke and suffering worse functional outcomes than White patients, including increased mortality and higher rates of post-stroke rehospitalization (Cruz-Flores et al., 2011). Racial disparities in stroke outcomes have been linked to suboptimal control of risk factors such as hypertension, lack of access to health care, and decreased utilization of neurologic services (Kulshreshtha et al., 2013; Saadi et al., 2017; Sacco et al., 2017). Stroke disparities may also be related to the quality of care delivered during the acute stroke episode (Kaufman et al., 2019). Black patients, for example, are more likely to have delayed arrival to emergency departments (ED) and longer ED wait times (Cruz-Flores et al., 2011). They are less likely to receive early thrombolysis or primary and secondary stroke prevention services (Boehme et al., 2014; Faigle et al., 2017).

Delays in appropriate care and subsequent disparate outcomes represent examples of structural racism, which is characterized by the alignment and allocation of health services in such a fashion that unfairly advantages and disadvantages members of some racial groups over others (Churchwell et al. 2020; National Academies of Science, Engineering, and Medicine [NASEM], 2017). Persistent disparities in ischemic stroke outcomes may also be linked to the pernicious effects of previously segregated health care facilitates and long-standing differences in the quality of care that minorities can access (Jones, 2000). Another aspect of structural racism that has not been fully explored is whether Black ischemic stroke patients are cared for in settings with insufficient nursing resources. Organizational characteristics, such as nurse staffing, represent modifiable aspects of health care systems that may be allocated to result in differential health outcomes. There is a substantial evidence base suggesting better outcomes when patients are cared for in settings with fewer patients per nurse (Aiken et al., 2002; Bruyneel et al., 2015), though recent studies suggest that Black patients are more often cared for in hospitals with lower quality of nursing care (Brooks Carthon et al., 2015).

Nurses are involved in nearly every aspect of stroke care, including acute medical stabilization, residual motor and neuropsychological dysfunction management, directing care coordination, and providing transitional care support (Camicia & Lutz, 2016). However, a growing body of research suggests that stroke outcomes may be compromised in hospitals with poor staffing levels. In their study of 185 hospitals, Cho and Yun (2009) found that patients cared for in better-staffed ICUs had lower odds of in-hospital mortality. Similarly, in their study of weekend staffing on 103 stroke units, Bray et al. (2014) found that units with better nurse staffing had lower odds of 30-day mortality. While the link between nurse staffing levels and stroke outcomes is relatively well established, none of the prior literature has examined this relationship among Black stroke patients raising questions about how staffing resource allocation influences outcome disparities.

Black patients are frequently hospitalized with more complex social needs (e.g., higher rates of poverty, housing insecurity, difficulties accessing health care) and higher rates of underlying comorbidities (Noonan et al., 2016; National Center for Health Statistics, 2016). Nurses are well-positioned to address social and clinical complexities arising during hospitalization; yet, a growing number of studies have found that they cannot address these concerns due to a lack of resources and excessive patient workloads (Brooks Carthon et al., 2019). If essential nursing care following ischemic stroke is omitted more frequently due to insufficient staffing levels in hospitals where minorities receive, then it is plausible that this may contribute to the disparities experienced by Black stroke patients.

Our interest in examining the relationship between nursing care and disparities in ischemic stroke outcomes led to examining stroke readmissions and nurse staffing. Over the past decade, studies by McHugh and Ma (2013) and others have found higher odds of readmissions in hospitals with poor nurse staffing (Lasater & McHugh, 2016). This study built on prior work and sought to determine if there are differences in prevalence of 7- and 30-day readmissions for Black ischemic stroke patients and examine whether these differences are associated with nurse staffing. This study’s findings will help inform interventions to address structural inequities that lead to persistent disparities in stroke outcomes.

Methods

Design

This cross-sectional study utilized three data sources that were linked using a common hospital identifier: (a) the 2015–2016 RN4CAST–US survey of registered nurses, (b) administrative patient discharge records (2015–2016), and (c) the American Hospital Association’s Annual Survey of hospitals (2015). The study included adult non-federal acute care hospitals in four states (California, Florida, New Jersey, and Pennsylvania) that participated in the 2015 American Hospital Association’s Annual Survey and 10 or more nurse respondents to the Registered Nurse Forcasting Study (RN4CAST–US) survey. Data from the RN4CAST–US survey includes the hospital name linked to the American Hospital Association’s Annual Survey. A common hospital identifier included in the patient discharge abstracts is then used to link the American Hospital Association’s Annual Survey and the RN4CAST–US survey to produce the final analytic data set. Our institution granted institutional review board approval.

Setting and Sample

Nurses

This study utilized survey data of nurses collected through the RN4CAST–US and used a modified Dillman sampling approach (Dillman et al., 2014). The survey was mailed to the homes of a random sample of 231,000 nurses holding an active RN license and yielded a 26% (n = 52,510) response rate (Lasater et al., 2019). Survey methodology details are published elsewhere (Sloane et al., 2018). To address concerns of nonresponse bias, investigators conducted a resurvey of initial nonresponders. This resurvey resulted in almost a 90% response rate, and when compared to the primary survey, no statistically significant differences were found, suggestive of no evidence of nonresponse bias (Lasater et al., 2019). The survey included items related to demographics, workload, and the work environment. Nurses were also asked to provide detailed information about their employing hospital (including hospital name), allowing responses across nurses to be aggregated to the hospital level. On average, there were 38.6 nurse respondents (range: 10–204) per study hospital.

Hospitals

We utilized the American Hospital Association’s Annual Survey and the RN4CAST–US survey to characterize the 542 study hospitals. The American Hospital Association’s Annual Survey was used to describe additional hospital characteristics, including organizational, structural, and financial measures that may be associated with nursing measures and patient outcomes. Information on Primary Stroke Center (PSC) Certification or Comprehensive Stroke Center (CSC) Certification was derived from the American Heart Association’s (AHA) website (AHA, 2020).

Patients

The sample of patients consisted of those aged 65 years and older with a primary diagnosis of ischemic stroke. Patient discharge data were obtained from state agencies, including California’s Office of Statewide Health Planning and Development, the New Jersey Department of Health and Senior Services, the Pennsylvania Health Care Cost Containment, and the Agency for Health Care Administration in Florida. Included in these data were patient demographics, discharge disposition, length of stay, primary and secondary diagnoses, and procedure codes based on the International Classification of Diseases 9th revision (ICD–9) Clinical Medication for admissions prior to October 1, 2015, and ICD–10 codes for admissions on and after October 1, 2015. Patients were included with a primary diagnosis of ischemic stroke using ICD–9 and ICD–10 codes (Supplemental Digital Content [SDC], Table 1). Patients were excluded if they were (a) not Black or White, (b) under the age of 65, (c) experienced a hemorrhagic stroke, and (d) transferred to another hospital, left against medical advice, or died while in the hospital (Yale New Haven Health Services Corporation/Center for Outcomes Research & Evaluation [YNHHSC/CORE], 2017). Our final analytic sample consisted of 98,150 patients in 542 hospitals (SDC, Figure 1).

Variables and Measures

Outcome Variable (Analyzed at the Patient Level)

The primary outcome variables were all-cause 7- and 30-day readmission from an index admission for ischemic stroke. The index admission for ischemic stroke was defined based on the Centers for Medicare and Medicaid Services (CMS) Risk-Standardization Readmission Measures (YNHHSC/CORE, 2017). The patient’s first admission during the study period was selected as the index. In the state databases, each patient is assigned a unique pseudo-identifier. Therefore, we were able to identify readmissions even when patients were readmitted to a different hospital.

Explanatory Variable (Analyzed at the Hospital Level)

Nurse staffing was our primary explanatory variable and derived from the RN4CAST–US survey. Each nurse surveyed reported the numbers of nurses and patients on the unit on the last shift worked. Like others, we created the staffing measure by dividing the average number of patients reported by nurses by the average number of nurses on the unit for that same shift (Bettencourt et al., 2020; Sloane et al., 2018). We then averaged these individual nurse reports across all nurses working in each hospital to estimate average hospital nurse staffing. Our measure of staffing has good predictive validity and has been repeatedly used in studies examining patient outcomes, including readmissions (Brooks Carthon et al., 2021).

Potentially Confounding Variables (Analyzed at the Hospital Level)

Prior studies have revealed an association between the nurse work environment and 30-day readmissions; hence we control for this in our analysis (Brom et al., 2021; McHugh & Ma, 2013). The nurse work environment was measured from the RN4CAST–US using the validated and reliable Practice Environment Scale of the Nursing Work Index (PES–NWI; Lake, 2007). The PES–NWI consists of 31 items; it has five subscales representing the domains of participation in hospital affairs, nurse manager leadership, foundations for quality of care, relations between nurses and physicians, and staffing and resource adequacy. We computed each subscale mean and averaged them across all nurses in each hospital to create hospital-level subscale scores and then averaged these subscale scores to compute a composite score (Lake, 2007). Hospitals were categorized based on those in the lowest 25th percentile (“poor”), the middle 50th percentile (“mixed”), and the top 25th percentile (“good”; Brom et al., 2021).

We included control variables for other potentially confounding structural hospital factors (e.g., teaching status, size, technology status, ICU staffing levels) associated with patient outcomes in prior work (Aiken et al., 2002). Hospital size was categorized by the number of beds (< 250, 251–500, and > 500 beds). Teaching status was categorized by the resident or fellow to bed ratio as major teaching (> 1:4 trainee-to-bed ratio), minor teaching (≤ 1:4 trainee-to-bed ratio), and nonteaching (no postgraduate trainees). High-technology hospitals were defined as those that performed open-heart surgery, major organ transplants, or both. We additionally included an indicator for the state and region (urban/rural) where hospitals were located. Our staffing measure is constructed from nurses in clinical units with different staffing requirements; hence we account for the proportion of nurses working on medical-surgical floors or ICUs. We accounted for CSC or PSC certifications status (AHA, 2020) due to association of CSC and PSC with better patient outcomes in prior studies (Man et al., 2017, 2020). These hospitals meet established standards, including acute stroke teams, specific diagnostic testing capabilities, and require continuing education.

Because differences in readmissions may also be more prevalent in hospitals caring for a higher proportion of low-income patients, we account for the proportion of Medicaid patients by creating a variable to indicate hospitals in the top 25th percentile of Medicaid percentage of inpatient days (Ferro et al., 2019). Finally, we risk-adjusted for other factors that might affect stroke outcome including 37 patient characteristics such as age, gender, transfer status, 29 of the Elixhauser comorbidities based on prior work and CMS Risk-Standardization Readmission Measures, payor type, length of stay, and whether the patient underwent mechanical thrombectomy (Elixhauser et al., 1998; YNHHSC/CORE, 2017; Zhong et al., 2016).

Analysis

First, patient, hospital, and nursing characteristics were compared between Black and White patients using chi-square tests for categorical variables and t-tests for continuous variables. Next, a series of multilevel, logistic regression models were employed to estimate the likelihood of 7- and 30-day readmissions before and after, controlling for confounding factors. All logistic regression models were estimated using Huber-White (robust) procedures to account for the clustering of patients within hospitals and adjust for the standard errors of the parameter estimates appropriately (Rogers, 1994). Four sequential models were constructed:

1. An unadjusted model examining the bivariate association between 7- and 30-day readmissions and race, and adjusted models which reestimate the race/readmission association after adjusting for

2. patient characteristics,

3. patient and hospital characteristics including nurse staffing, and

4. all of the main effects of the characteristics in the third model plus interaction between race and nurse staffing centered at the mean.

All analyses were conducted using Stata 15 (Stata Statistical Software: Release 15. StataCorp. 2017. College Station, TX: StataCorp LLC.).

Results

Sample Characteristics

Our analysis included 98,150 patients who experienced an ischemic stroke in 542 hospitals. Table 1 displays characteristics of Black and White patients. Of these patients, 85,878 (87.5%) were White and 12,272 (12.5%) were Black. A higher percentage of Black patients than White patients experienced a 7-day (4.8% vs. 3.8%, p < .001) and 30-day readmission (12.7% vs. 10.0%, p < .001). Black patients were younger, less frequently male, and less frequently insured by Medicare. Black patients experienced a longer length of stay (5.6 vs. 4.4 days, p < .001) and had more comorbidities (3.4 vs. 3.0, p < .001) compared to White ischemic stroke patients. Black patients were also discharged to skilled nursing facilities more often (30.6% vs. 27.4%, p < .001) than White ischemic stroke patients.

Table 1:

Patient Characteristics n= 98,150

Characteristic	White (87.5%, n = 85,878)	Black (12.5%, n = 12,272)	p-value
Readmission within 7 days, n (%)	3,247 (3.8)	586 (4.8)	< .001
Readmission within 30 days, n (%)	8,635 (10.0)	1,555 (12.7)	< .001
Length of stay in days, mean (SD)	4.4 (5.9)	5.6 (7.0)	< .001
Age, mean (SD)	80.0 (8.5)	76.4 (8.1)	< .001
Male, n (%)	38,938 (45.3)	5,064 (41.3)	< .001
Insurance Type, n (%)			< .001
Medicare	79,722 (92.9)	10,625 (86.8)
Medicaid	381 (0.4)	349 (2.8)
Private	4,703 (5.5)	952 (7.8)
Self-pay	153 (0.2)	90 (0.7)
No Charge	19 (0.02)	14 (0.1)
Other	821 (0.96)	204 (1.7)
Transferred, n (%)		423 (3.4)	.019
Discharge destination, n (%)			< .001
Home	23,893 (27.8)	3,096 (25.2)
Skilled nursing facility	23,548 (27.4)	3,752 (30.6)
Home with home health care	13,817 (16.1)	2,161 (17.6)
Intermediate care	440 (0.5)	48 (0.4)
Other facility	24,130 (28.1)	3,210 (26.2)
Mechanical thrombectomy, n (%)	198 (0.23)	25 (0.20)	.559
Number of comorbidities, mean (SD)	3.0 (1.7)	3.4 (1.7)	< .001
10 most common comorbidities, n (%)
Hypertension	74,505 (86.8)	11,548 (94.1)	< .001
Diabetes w/o chronic complications	18,146 (21.1)	3,987 (32.5)	< .001
Fluid & electrolyte disorders	18,086 (21.1)	3,089 (25.2)	< .001
Hypothyroidism	17,744 (20.7)	1,034 (8.4)	< .001
Renal failure	14,617 (17.0)	3,193 (26.0)	< .001
Chronic pulmonary disease	14,497 (16.9)	1,739 (14.2)	< .001
Congestive heart failure	13,016 (15.2)	2,178 (17.7)	< .001
Valvular disease	13,025 (15.2)	1,172 (9.5)	< .001
Deficiency anemias	11,138 (13.0)	2,577 (21.0)	< .001
Peripheral vascular disease	10,988 (12.8)	1,611 (13.1)	.303

Note. Comorbidities determined by Elixhauser

Table 2 shows differences for Black and White patients for the hospital and nursing characteristics where they received care. Black patients were more likely than White patients to be cared for in hospitals with lower nurse staffing ratios (4.3 and 4.4, respectively; p < .001). Black patients were also more likely than White patients to be cared for in large urban teaching hospitals with high technology, stroke center designation, and a higher percentage of Medicaid beds. Hospitals with the best staffing levels were also those with better work environments, large high-technology stroke centers, and a higher percentage of Medicaid beds (SDC, Table 2).

Table 2:

Hospital and Nursing Characteristics

Characteristic	Hospital (n = 542)	White (87.5%, n = 85,878)	Black (12.5%, n = 12,272)	p-value
Patients per nurse, mean (SD)	4.4 (1.0)	4.4 (0.8)	4.3 (0.8)	< .001
Categorical staffing, n (%)				< .001
< 3 patients/nurse	36 (6.6)	2,869 (3.3)	862 (7.0)
3–3.99 patients/nurse	165 (30.4)	24,872 (29.0)	3,513 (28.6)
4–4.99 patients/nurse	210 (38.7)	41,690 (48.5)	5,906 (48.1)
5–5.99 patients/nurse	93 (17.2)	13,254 (15.4)	1,594 (13.0)
≥ 6 patients/nurse	38 (7.0)	3,193 (3.7)	397 (3.2)
Work environment, mean (SD)	2.8 (0.2)	2.8 (0.2)	2.8 (0.2)	.004
Cat. work environment, n (%)				< .001
Poor	136 (25.1)	18,521 (21.6)	2,421 (19.7)
Mixed	271 (50.0)	42,174 (49.1)	6,240 (50.8)
Best	135 (24.9)	25,183 (29.3)	3,611 (29.4)
Bed size, n (%)				< .001
Small	45 (8.3)	2,956 (3.4)	162 (1.3)
Medium	227 (41.9)	25,478 (29.7)	2,830 (23.1)
Large	270 (49.8)	57,444 (66.9)	9,280 (75.6)
Teaching status, n (%)				< .001
None	222 (42.3)	36,997 (43.9)	3,360 (28.1)
Minor	254 (48.4)	38,877 (46.2)	6,044 (50.5)
Major	49 (9.3)	8,313 (9.9)	2,561 (21.4)
High technology, n (%)	280 (52.6)	56,443 (66.7)	8,333 (69.1)	< .001
Stroke Center, n (%)	31 (5.7)	8,454 (9.8)	1,768 (14.4)	< .001
Urban, n (%)	535 (98.7)	85,411 (99.5)	12,252 (99.8)	< .001
Percent Black stroke patients, mean (SD)	14.5 (18.2)	9.8 (10.9)	31.6 (25.6)	< .001
High Medicaid, n (%)	149 (27.5)	14,386 (16.7)	3,519 (28.7)	< .001

Odds of 7-and 30-Day Readmissions for Black and White Patients

Table 3 displays the results of the multilevel, logistic regression models. In the unadjusted Model 1, the odds of 7- and 30-day readmission are higher for Black ischemic stroke patients than for White patients by a factor of 1.28 (95% CI [1.15, 1.41] and 1.30 (95% CI [1.21, 1.40]), respectively. In our fully adjusted models accounting for patient and hospital characteristics in Model 3—the odds ratios (OR) reflecting the Black–White difference in 7-day and 30-day readmissions are 1.11 (95% CI [1.01, 1.24]) and 1.13 (95% CI [1.05, 1.22]), respectively. Model 4 reveals an interaction between race and nurse staffing that is both statistically significant and substantively important. The main effect of race in Model 4 remains pronounced for both 7-day (OR = 1.12; 95% CI [1.01, 1.24]) and 30-day readmissions (OR = 1.14; 95% CI [1.07, 1.23]. However, the interaction term (race * nurse staffing) in the models for 7-day (OR = 1.11; 95%CI [1.01, 1.22]) and 30-day readmissions (OR = 1.10; 95% CI [1.02, 1.18]) demonstrates that the Black–White difference is more pronounced when patient-to-nurse ratios are higher (i.e., when staffing is poorer). Though, the staffing effect is decidedly smaller for White patients (OR = 0.99) than for Black patients (OR = 0.99 × 1.11 = 1.10) for 7-day readmission and was similar for 30-day readmission.

Table 3:

Sequential Modeling Indicating the Effects of Racial Disparities and Nurse Staffing on 7- and 30-Day Readmissions Among Stroke Patients

	Unadjusted OR (95% CI)	Adjusted for patient characteristics OR (95% CI)	Adjusted for patient and hospital characteristics OR (95% CI)	Fully adjusted with interaction of centered staffing and race OR (95% CI)
7-Day readmission
Black	1.28*** (1.15–1.41)	1.17** (1.05–1.30)	1.11* (1.01–1.24)	1.12* (1.01–1.24)
Staffing			1.01 (0.95–1.07)	0.99 (0.93–1.05)
Interaction			--	1.11* (1.01–1.22)
30-day readmission
Black	1.30*** (1.21–1.40)	1.17*** (1.09–1.26)	1.13** (1.05–1.22)	1.14*** (1.07–1.23)
Staffing			1.02 (0.98–1.06)	1.01 (0.96–1.04)
Interaction			--	1.10* (1.02–1.18)

Note. All models utilize multivariable logistic regression that accounts for clustering within hospitals.

^*p < .05

^**p < .01

^***p < .001

Predictive Odds of Readmission at Each Staffing Level

Figures 1 and 2 illustrate how 7- and 30-day readmission differs for Black and White patients across hospitals at each level of nurse staffing and reveal how the effect of varying staffing levels is more pronounced for Black than White patients. For example, in hospitals in which nurses care for the average amount of patients (four per nurse), the odds of 30-day readmission are 0.14 for Black patients (95% CI [0.13, 0.15]) and 0.11 for White patients (95% CI [0.11, 0.12]). This racial disparity narrows when the nurse cares for fewer patients such that when nurses care for an average of two patients, the odds of 30-day readmission are 0.11 for Black patients (95% CI [0.10, 0.13]) and 0.11 for White patients (95% CI [0.10, 0.12]). Conversely, this disparity widens when nurses care for more patients. In hospitals where nurses care for seven patients, the odds of 30-day readmission are 0.19 for Black patients (95% CI [0.15, 0.22]) and 0.11 for White patients (95% CI [0.10, 0.13]).

Figure 1:

Odds on 7-day readmission for Black and Whites, in hospitals with different staffing levels

Figure 2:

Odds 30-day readmission for Black and Whites, in hospitals with different staffing levels

Discussion

Our study is the first to demonstrate an association between nurse staffing and 7- and 30-day readmission disparities for Black ischemic stroke patients. Black ischemic stroke patients were readmitted at higher levels at both 7- and 30-days following discharge. We note that readmission disparities are more prominent in poorly staffed hospitals than better-staffed hospitals. These disparities are virtually eliminated in settings where the most intense nursing resources are in place. The association between staffing levels and 7-day readmission disparities is particularly notable since these early rehospitalizations are believed to be more closely related to care received during the acute stay.

Our findings have important implications for nursing care. Nurses play a vital role in addressing the clinical needs of stroke patients by developing discharge care plans and identifying and communicating barriers that may impede successful transitions (Camicia & Lutz, 2016). While we found no direct effect of staffing on readmissions for White stroke patients, variations in staffing appear to have a differential effect on Black stroke patients. For example, in hospitals in which nurses care for the average amount of patients (four per nurse), the odds of 30-day readmission are 14% for Black patients and 11% for White patients. This suggests that while Black and White ischemic stroke patients are on average hospitalized in settings with similar nurse-to-patient ratios (4.3 vs. 4.4), these levels may not be sufficient to meet the clinical needs of Black ischemic stroke patients. We note, for example, that the racial disparity in 30-day readmissions is virtually eliminated when Black ischemic stroke patients are cared for in settings with the most robust staffing levels. In these settings, the odds of 30-day readmission are 11% for both Black and White patients, respectively. This finding suggests that efforts to address structural inequities in stroke outcomes should not simply attempt to ensure that staffing levels for Black patients are equal to those of White patients. Instead, efforts to attain more equitable outcomes may require increased staffing levels beyond that of White patients to ensure that staffing resources are sufficient to meet population-specific clinical needs.

Other studies have similarly noted the pronounced effect of nurse staffing on the outcomes of hospitalized Black patients. In a recent study of in-hospital cardiac arrest across 75 U.S. hospitals, Brooks Carthon et al. (2021) found that each additional patient cared for by a nurse lowered the odds of survival for Black patients by 8% but only by 3% for White patients. Similarly, in their examination of 30-day readmissions among older hip and knee replacement patients, Lasater and McHugh (2016) found that each additional patient that nurses cared for increased the odds of readmission by 15% for Black patients, compared to an 8% increase among White patients.

We suspect that Black stroke patients may be susceptible to lower staffing levels due to increased clinical complexity, preexisting medical comorbidities, or other social factors, such as fewer economic resources and connections to community-based services, some of which may not be fully accounted for in our analyses. In our sample, Black ischemic stroke patients were more frequently admitted to a skilled nursing facility (SNF) following discharge, suggesting the need for extended skilled nursing care due to functional decline and physical deconditioning. For Black ischemic stroke patients not discharged to SNF, the need for additional support may also be warranted. In their 2019 study, Magwood et al. (2019) found that Black stroke patients and their caregivers expressed a desire for more social support and resources during their stroke recovery. They also noted barriers to recovery, including a lack of health education and care coordination. Due to their proximity to patients and families and training in holistic care, acute care nurses are well-positioned to address hospitalized stroke patients’ clinical and social needs, yet, organizational constraints may restrict their ability to do so. In a qualitative study based in an acute care setting, nurses were asked to reflect on how social needs were assessed, and once discovered, how they were prioritized and incorporated into goal setting and discharge planning. Nurse participants reported feeling limited in their ability to address hospitalized patients’ social concerns, partly due to excessive workloads and competing job demands (Brooks Carthon et al., 2019). Efforts to address readmission disparities among Black ischemic stroke patients will require culturally informed initiatives inclusive of chronic disease management, discharge education, attention to the social determinants of health, and the provision of psychosocial support. NASEM’s recently released Future of Nursing report addresses the role of nurses in meeting the clinical and social needs of patients though raises concerns with the current level of institutional support, noting,

If nurses and other clinicians are expected to fulfill their professional mandates for social justice, including addressing complex social needs and to spend the time and cognitive effort to address both health and social issues, system-wide changes will be needed to provide space and support for that shift. (NASEM, 2021, p. 337)

Health care systems have indeed invested in a range of programs and interventions to reduce readmissions, such as discharge planners and transitional care programs; the results of these efforts have been mixed and difficult to scale (Condon et al., 2016; Facchinetti et al., 2020). In contrast, our findings suggest that ensuring sufficient system-level resources through adequate staffing levels may facilitate more intensive nursing care and engagement with the clinical and psychosocial needs of Black ischemic stroke patients—resulting in fewer readmissions and better overall outcomes.

Limitations

Our study is limited to using available claims data, which allowed us to risk adjust for various patient factors, though it did not allow us to account for stroke severity. We were, however, able to account for many medical conditions, and in prior studies, stroke severity did not explain disparities in stroke outcomes (Faigle et al., 2017). We also note that our use of the 2015 American Hospital Association’s Annual Survey limits our ability to account for the role of other health care providers caring for stroke patients, such as occupational or physical therapists, or differences in the quality of care delivered by other providers, including neurologists or intensivists. Additionally, we acknowledge that other factors may result in readmissions for minority stroke patients, such as older age and hospital stroke volume (Zhong et al., 2016). These factors are not easily modifiable by hospitals or health care providers. We note that nurse staffing levels can provide only limited detail about the actual care provided to patients during the acute stroke episode. The quality of thrombolytic programs, for instance, is not measured in this study, nor are other factors such as care coordination or additional transitional care support. Though, these initiatives may also influence stroke outcomes. We are also constrained by our race variable, which was determined from state discharge abstracts; we are vulnerable to race misattribution when not self-reported. Often racial identity serves as a proxy for the effect of systemic racism and discrimination, and future studies should consider more direct measures of racism and discrimination to better examine root causes of inequities in health outcomes. Finally, while we acknowledge that variations in staffing levels may reflect systemic inequities long endured by historically marginalized patients, we are unable to account for personally mediated racism, discrimination, or implicit biases held by nurses or other health care providers, factors which may also be linked to poor outcomes among minority patients.

Conclusion

Disparities in 7- and 30-day readmissions among ischemic stroke patients may be linked to the level of nurse staffing in the hospitals where they receive acute stroke care. Addressing long-standing systemic inequities will require systemwide investments to raise staffing levels that allow nurses to sufficiently address the clinical and social needs of Black ischemic stroke patients.