Study Population

From the NIS data,¹⁶ 27,135 adult patients at least 18 years of age, identified by International Classification of Diseases, Ninth Revision (ICD-9) and International Statistical Classification of Diseases, Tenth Revision (ICD-10) codes for TTS (ICD-9 429.83; ICD-10 I51. 81) and admitted to hospital between 2012 and 2016, were included in this study. The final sample consisted of 2,317 Black patients and 24,817 White patients.

Baseline demographic characteristics—such as age, sex, body mass index, and year of hospital admission—and clinical characteristics—such as smoking status; presence of diabetes, hypertension, or hyperlipidemia; and a history of myocardial infarction (MI), percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG), atrial fibrillation or flutter, peripheral artery disease, and sepsis—,were identified based on ICD-9 and ICD-10 codes. In-hospital outcomes such as length of stay, cardiogenic shock, cardiac arrest, cerebrovascular accidents, invasive mechanical ventilation, mechanical circulation support, acute kidney injury, and mortality were registered using the ICD-9-Clinical Modification codes.¹⁷

Statistical Analysis

Given the observed baseline covariates, a logistic regression model was constructed to calculate the predicted probability and propensity score of being Black for each patient at the time of admission. The propensity score represented the relationship between prespecified baseline characteristics (ie, sex; age; smoking status; presence of diabetes, hypertension, or hyperlipidemia; and history of MI, PCI, CABG, atrial fibrillation or flutter, peripheral artery disease, or sepsis) and being Black as a single characteristic. A matching approach was then used to obtain a matched cohort with a similar propensity score between Black patients and White patients to balance observed covariates.

The propensity score distribution between the 2 groups in the original sample and the matching sample was compared to assess the balance in the matched cohort that the propensity model produced. The standardized difference was calculated as the mean difference between the 2 groups divided by a measure of the SD of each covariate.

Baseline demographic and clinical characteristics were reported as a medians for continuous variables and percentages for categorical variables. A Wilcoxon rank sum test and a χ² test were conducted for continuous and binary variables, respectively, to compare the proposed groups (Black patients and White patients) before and after matching. In-hospital complications and inpatient mortality were compared between the matched groups to determine differences in clinically significant outcomes between Black and White patients.

The logistic regression model was used to calculate the odds ratio of in-hospital mortality for the combined race-sex effect, using a group of White women as the reference in the matched group after adjusting for baseline characteristics (sex; age; smoking status; presence of diabetes, hypertension, or hyperlipidemia; and history of MI, PCI, CABG, atrial fibrillation or flutter, peripheral artery disease, or sepsis). Statistical analyses were performed using SAS, version 9.4 (SAS Institute Inc) and Stata, version 16 (StataCorp LLC) statistical software. All tests used α = .05 as the probability for a type I error.

Unmatched Population

The total number of patients admitted with TTS between 2012 and 2016 was 27,135 (median age, 68 years; 86% women). There were 2,317 (8.54%) Black patients and 24,817 (91.46%) White patients. The group of Black patients was younger (63 vs 68 years of age, P< .01) and had a lower percentage of women (81% vs 86%, P < .01). Black patients were also more likely to have diabetes (31.46% vs 20.81%, P < .01) and hypertension (54.9% vs 54.5%, P = .01) than their White counterparts. Conversely, hyperlipidemia (39.7% vs 45.4%, P = .01) and atrial fibrillation (13.5% vs 21%, P < .01) were more frequent in White patients (Table I). Black patients had more in-hospital complications; this finding did not change after adjustment for age and sex (Supplementary Table I).

TABLE I. Differences in Demographic Characteristics Between Black and White Patients With TTS Before Matching

View Fullscreen

Matched Population

Propensity score distribution between the 2 groups in the original sample and the matched sample was used to assess the balance between baseline confounding factors before and after matching. The propensity score distribution differed between the 2 groups in the original sample (Fig. 1) but was similar in the matched sample (Fig. 2). Figure 3 shows that the 2 groups' age distribution at admission was similar after matching. Figure 4 shows a reduction in the standardized differences in propensity score in the matched sample compared with the original sample. The standardized differences in the matched sample were less than 5%.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

After matching, 4,628 patients were identified with a similar propensity score between Black patients (n = 2,314) and White patients (n = 2,314). Sex and age distributions were identical between the 2 groups. The 2 groups also had similar baseline characteristics, including CV risk factors (Table II). The difference in in-hospital complications between Black patients and White patients is reported in Table III. The prevalence rates of cerebrovascular accidents (P < .01), length of stay (P < .01), and acute kidney injury (P < .01) were significantly higher in Black patients. The rate of in-hospital mortality was still significantly higher in Black patients than in White patients (6.1 vs. 4.5, P < .01) (Table III).

TABLE II. Differences in Demographic Characteristics Between Black and White Patients With TTS After Matching

View Fullscreen

TABLE III. Differences Between Black and White Patients With TTS for In-Hospital Complications After Matching

View Fullscreen

Logistic Regression Analysis

A logistic regression analysis in the unmatched groups, adjusted for baseline characteristics, showed a 25% increased mortality rate in Black patients vs White patients (odds ratio, 1.25 [95% CI, 1.03-1.51]; P = .02). When sex and race were combined and adjusted for CV risk factors, Black men had the highest mortality rates, followed by White men and Black women (Supplementary Table II). The matched groups showed the same results when sex and race were combined: Black men had the highest mortality rates, followed by White men and Black women (Table IV).

TABLE IV. Risk of Inpatient Mortality Among Matched Groups for Combined Sex and Race^a

View Fullscreen

The Uniqueness of the Present Study

The present study represents one of the most extensive reports on Black patients admitted to US hospitals with TTS. Key features to be highlighted are (1) Black patients had an overall greater occurrence of CV risk factors and in-hospital outcomes, including length of hospital stay and mortality rates; (2) Black patients had higher in-hospital mortality rates after balancing the study groups by propensity score matching, although CV risk factors were similar between the 2 groups; and (3) when race and sex were combined, sex likely masked the higher susceptibility to mortality in Black men.

In the unmatched groups, the higher prevalence of CV risk factors in Black patients could be considered the predisposing factor for the worst outcome.^25–27 In particular, Black men tend to develop hypertension earlier in life than Black women and White individuals, with the onset of related complications starting at a younger age.^6,26 Diabetes is common in patients with TTS, particularly men, and is frequently associated with hypertension.²⁸ Diabetes is also 1 of the main predictors of short-term and long-term mortality in the InterTAK Prognostic Score¹⁵ and in other studies^28,29 for patients with TTS. The 2 racial groups were similar, however, in their rates of prior MI and history of PCI and CABG, as reported by others.^13,30 In terms of in-hospital complications, Black patients presented with higher rates of cerebrovascular accident, acute kidney injury, and inpatient mortality and had longer hospital stays than their White counterparts,^18,21 suggesting that race may play a role in the outcomes of Black patients with TTS; this finding is in contrast with those of previous studies.^18,19,31

Propensity score matching was used to create 2 balanced groups and overcome the confounding baseline factors between them. The 2 groups were similar in age, sex, CV risk factors, and incidence of CVD, but Black patients still had higher rates of in-hospital complications. When sex and race were combined, mortality rates were significantly higher in Black men, followed by White men and Black women. It is likely that male sex masks a baseline predisposition among Black individuals for the worst outcome. The mechanisms underlying racial differences in TTS outcomes remains to be clarified. Differences in genes that control sympathetic activity, however, and differences in endothelium homeostasis have been identified in Black patients. These distinctive characteristics are associated with increased sympathetic tone, leading to an aggravated CV response to physiologic and psychological stimuli in Black patients compared with White patients.³² Acetylcholine testing could be applied in this setting to highlight the endothelial dysfunction hypothesized to be particularly reactive in Black patients.³³ This information, however, is not part of the NIS registry. Male sex characteristics, such as hormone status,³⁴ different gene expressions related to estrogens,³⁵ a higher inflammatory state, whether a physical stressor was involved,³⁶ and higher susceptibility to develop CVD and complications at a younger age³⁷ could explain the poorer outcomes in Black men.

Limitations

Study limitations intrinsic to the NIS data set must be addressed. The NIS data set is a retrospective report based on administrative data using ICD-10 and ICD-9 coding. Consequently, the accuracy of diagnosis relies entirely on proper coding. The NIS database also captures hospitalizations rather than individual patients. The actual prevalence of TTS in Black patients may therefore need to be considered. Laboratory data (including cardiac biomarkers), electrocardiographic data, and echocardiographic data were also unavailable. Predischarge evaluations for echocardiographic recovery, long-term prognosis information, and prehospital admission events were not provided and could not be considered in the study. Specific diagnostic tests, such as acetylcholine testing for endothelial dysfunction, were not reported and likely not performed in general hospitals. The NIS database registers only in-hospital events, so no prehospital admission data on complications and mortality in general and in events related to TTS in particular were available.