The Association Between Hyperlipidemia and In-Hospital Outcomes in Takotsubo Cardiomyopathy.

PURPOSE: Hyperlipidemia (HLD) is one of the most common cardiovascular risk factors and is prevalent in patients with takotsubo cardiomyopathy (TCM), but the association between HLD and TCM patients' outcomes is unclear. We investigated the impact of HLD on the in-hospital outcomes of TCM patients. PATIENTS AND METHODS: Our retrospective cohort study used the latest available data from the National Inpatient Sample (2016-2017). Using the ICD-10 code, we identified 3139 patients with a primary diagnosis of TCM, 1530 of whom had HLD. We compared in-hospital outcomes between HLD and non-HLD groups before and after propensity score matching.
RESULTS: In the unmatched cohort, the HLD group had lower incidences of cardiac arrest, cardiogenic shock, and acute respiratory failure (ARF); shorter length of stay (LOS); and lower total charges (All p<0.05). In-hospital mortality (p=0.102) and ventricular arrhythmia (p=0.235) rates did not differ. After propensity score matching, the HLD group had lower rates of in-hospital mortality (1.1% vs 2.4%, p=0.027), ARF (9.1% vs 12.1%, p = 0.022) and cardiogenic shock (3.4% vs 5.6%, p=0.012), shorter LOS (3.20 ± 3.27 days vs 3.57 ± 3.14 days, p=0.005), and lower total charges (p=0.013). The matched groups did not differ significantly regarding cardiac arrest (p=0.141), ventricular arrhythmia (p=0.662) or acute kidney injury (AKI) (p = 0.167).
CONCLUSION: Counterintuitively, HLD was associated with better in-hospital outcomes in both the unmatched and propensity-matched cohorts of hospitalized TCM patients. Further studies are needed to investigate the mechanisms that may contribute to the association in TCM patients with HLD.

Introduction

Since the diagnosis of takotsubo cardiomyopathy (TCM) was introduced in the 1990s,1 there is much yet to be discovered about the disease. TCM is more prevalent among females, occurring around 10 times more often in females than males,2 and has similar clinical features to ACS, including typical ischemic chest pain, typical troponin and ECG changes.3 Once thought a rare and benign diagnosis (due to its self-limiting clinical course4,5), TCM, characterized by acute left ventricular systolic dysfunction without acute coronary artery obstruction, is increasingly recognized6,7 as having a substantial risk of mortality, similar to acute coronary syndrome (ACS).8

Hyperlipidemia (HLD) is a common metabolic disease in the United States, afflicting about 33.5% adults over age 20.9 HLD is characterized by high levels of low-density lipoprotein cholesterol (LDL-C) or high triglycerides (TG), and plays an important role in the pathogenesis of cardiovascular diseases.10,11 Previous studies have found that HLD was prevalent in TCM patients12–14 – a study by Summers et al investigating the risk factors of TCM found that 52% of TCM patients had HLD.12

In studying HLD’s associations with in-hospital outcomes of cardiovascular disease,15–17 findings counterintuitively suggest that low serum cholesterol level has a negative association with short-term outcomes. In Reddy et al’s study, lower levels of LDL-C were associated with a higher in-hospital mortality in patients with acute myocardial infarction.16 Another study indicated that patients with lower levels of total cholesterol were associated with higher in-hospital mortality in acute heart failure patients.17 While one study found the HLD did not affect the long-term mortality with TCM patients,18 the impact of HLD on the short-term outcomes of TCM patients has yet to be reviewed.

In this study, we sought to identify the association between HLD and the in-hospital outcomes among patients who were admitted for TCM using the latest available data from the National Inpatient Sample (NIS).

Method

Data Source

The NIS database is a stratified sample of 20% of all inpatient hospitalizations in the United States that includes data from >7 million unweighted and an estimated >35 million weighted hospital stays each year.19 The NIS database has been extensively used in many fields and it has the advantage of large sample sizes, and it contains patient characteristics such as a primary diagnosis for admission, comorbidities, and detailed patient demographics. As the information on NIS is de-identified, our study did not require the Institutional Review Board's approval. All comorbidities and inpatient complications are selected by the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) coding system from the NIS database. Patients admitted with HLD were identified by using ICD-10-CM codes E78.0–E78.5, which were consistent with the methodology used in previous studies.20 The ICD-10-CM codes used in this study are shown in Table 1.

Table 1

International Classification of Disease, 10th Edition, Clinical Modification (ICD-10-CM) Codes Used for Comorbidities and Complications

Comorbidities/Complications	ICD-10 Code
TCM	I5181
HLD	E780-E785
Comorbidities
Smoking	F17-F17299, Z720, Z87891
Hypertension	I10, I15-I159
DM	E10-E109, E11-E119, E13-E139
Obesity	Z683-Z6839, Z684-Z6845, E66-E669
Anxiety	F064, F40-F409, F41-F419
Depression	F32-F329, F33-F339, F341
COPD	J41-J418, J42, J43-J439, J44-J449
OSA	G473-G4739
CKD	N18-N189
PAD	I70, I71, I72, I73, I74, I77
Sepsis	A021, A227, A267, A327, A40-A409, A41-A419, A427, A5486, B377, O0337, O0387, O0487, O0737, O0882, O85, P36-P369, R652-R6521
Complications
Cardiac arrest	I46-I469
Cardiogenic shock	R570
Ventricular arrhythmia	I472, I490-I4902
AKI	N17-N179
ARF	J960-J9622

Abbreviations: AKI, acute kidney injury; ARF, acute respiratory failure; CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; HLD, hyperlipidemia; OSA, obstructive sleep apnea; PAD, peripheral artery disease; TCM, takotsubo cardiomyopathy.

Study Population

Using the ICD-10-CM code of I5181, we selected patients with a primary diagnosis of TCM from January 1, 2016, to December 31, 2017. Patients without discharge status were excluded. The patient selection process in this study is shown in Figure 1.

Figure 1

Flow chart of the selection process for the final patient sample used in this study.

Variables

The variables in our study include patient demographics (age, sex, race, household income, primary payer for the hospitalization) and hospital demographics (hospital type, hospital size and region). To further reduce selection bias, we included common cardiovascular comorbidities, such as smoking, hypertension, obstructive sleep apnea (OSA), peripheral artery disease (PAD), and other reported risk factors for TCM including chronic obstructive pulmonary disease (COPD),21,22 anxiety23 and depressive disorders,18 obesity,24 diabetes mellitus (DM),25 chronic kidney disease (CKD)26 and sepsis.27

In-Hospital Outcomes

The primary outcomes of our study were in-hospital mortality, length of stay (LOS) and total hospital charges. We also examined in-hospital complications including cardiac arrest, cardiogenic shock, ventricular arrhythmia, acute kidney injury (AKI), and acute respiratory failure (ARF).

Statistical Analysis

Continuous variables were expressed as a mean with standard deviation (SD) and were tested with a t-test, while categorical variables were expressed as percentages and tested with a chi-square test. We compared patients admitted for TCM with HLD and those without HLD prior to and after propensity score matching. The characteristics we selected to adjust for comparison were age, sex, race, and cardiovascular risk factors as described above.

We identified 3139 patients with a primary diagnosis of TCM with available discharge status by ICD-10 code. Based on whether the patient had a comorbidity of HLD, we divided patients with a primary diagnosis of TCM into two groups: one with documented comorbidity of HLD, the other one without. There are 1530 and 1609 patients in the HLD group and non-HLD group, respectively. Both groups had 1162 patients after propensity score matching.

The propensity score matching model was used to match selected baseline characteristics. Both groups of patients were matched into each group in a 1:1 target ratio. Finally, we compared the in-hospital outcomes between the two groups prior to and after adjustment. Differences were considered statistically significant if p < 0.05. All statistical analyses were performed by the R statistics software (version 3.6.1, R Development Core Team).

Results

Baseline Characteristics

Prior to matching, patients in the HLD group were older (68.71 ± 11.04 vs 64.73 ± 13.80, p < 0.001) compared to the non-HLD group. The HLD group had a higher proportion of certain comorbidities, such as hypertension (57.6% vs 40.8%, p < 0.001), DM (27.5% vs 12.6%, p < 0.001), obesity (15.4% vs 10.1%, p < 0.001), depressive disorders (20.1% vs 15.4%, p = 0.001), OSA (7.1% vs 3.9%, p < 0.001), CKD (10.7% vs 6.8%, p < 0.001) and PAD (8.8% vs 5.0%, p < 0.001). After propensity score matching, the baseline characteristics were comparable (p>0.05). Details of baseline characteristics of patients are shown in Table 2. We used standardized mean difference to examine the balance of the covariate distribution between HLD group and non-HLD group. After propensity score matching, all standardized mean difference is less than 0.1 between matched cohorts28 (Figure 2).

Table 2

Baseline Characteristics

Variables	Unmatched Cohort			Propensity-Matched Cohort
	TCM without HLD	TCM with HLD	P value	TCM without HLD	TCM with HLD	P value
n	1609	1530		1162	1162
Age, (mean (sd))	64.73 (13.80)	68.71 (11.04)	<0.001	67.63 (12.68)	67.46 (11.00)	0.719
Sex, n (%)			0.175			1
Male	163 (10.1)	135 (8.8)		100 (8.6)	100 (8.6)
Female	1444 (89.7)	1395 (91.2)		1062 (91.4)	1062 (91.4)
Unknown	2 (0.1)	0 (0.0)		0 (0.0)	0 (0.0)
Race, n (%)			0.192			0.865
White	1259 (78.2)	1235 (80.7)		930 (80.0)	950 (81.8)
Black	106 (6.6)	100 (6.5)		67 (5.8)	70 (6.0)
Hispanic	100 (6.2)	76 (5.0)		67 (5.8)	60 (5.2)
Asian/Pacific Islander	22 (1.4)	23 (1.5)		17 (1.5)	17 (1.5)
Native American	14 (0.9)	4 (0.3)		4 (0.3)	4 (0.3)
Other	35 (2.2)	32 (2.1)		26 (2.2)	20 (1.7)
Unknown	73 (4.5)	60 (3.9)		51 (4.4)	41 (3.5)
Patient location, n (%)			0.539			0.996
“Central” counties of metro areas of ≥1 million population	367 (22.8)	354 (23.1)		273 (23.5)	263 (22.6)
“Finge” counties of metro areas of ≥1 million population	370 (23.0)	394 (25.8)		280 (24.1)	292 (25.1)
Counties in metro areas of 250,000–999,999 population	411 (25.5)	357 (23.3)		279 (24.0)	278 (23.9)
Counties in metro areas of 50,000–249,999 population	168 (10.4)	158 (10.3)		122 (10.5)	120 (10.3)
Micropolitan counties	165 (10.3)	156 (10.2)		121 (10.4)	122 (10.5)
Non metropolitan or micropolitan counties	121 (7.5)	107 (7.0)		84 (7.2)	83 (7.1)
NA	7 (0.4)	4 (0.3)		3 (0.3)	4 (0.3)
Mean household income, n (%)			0.198			0.809
$1-$42,999	401 (24.9)	348 (22.7)		276 (23.8)	270 (23.2)
$43,000-$53,999	429 (26.7)	384 (25.1)		311 (26.8)	292 (25.1)
$54,000-$70,999	406 (25.2)	432 (28.2)		296 (25.5)	320 (27.5)
$71,000 or more	351 (21.8)	350 (22.9)		265 (22.8)	266 (22.9)
Unknown	22 (1.4)	16 (1.0)		14 (1.2)	14 (1.2)
Primary payer, n (%)			<0.001			0.998
Medicare	858 (53.3)	1009 (65.9)		715 (61.5)	718 (61.8)
Medicaid	175 (10.9)	99 (6.5)		86 (7.4)	88 (7.6)
Private including HMO	461 (28.7)	367 (24.0)		312 (26.9)	304 (26.2)
Self-pay	71 (4.4)	30 (2.0)		29 (2.5)	30 (2.6)
No charge	5 (0.3)	1 (0.1)		1 (0.1)	1 (0.1)
Other	39 (2.4)	23 (1.5)		19 (1.6)	21 (1.8)
Unknown	0 (0.0)	1 (0.1)		0 (0.0)	0 (0.0)
Hospital type, n (%)			0.438			0.863
Rural	96 (6.0)	78 (5.1)		69 (5.9)	64 (5.5)
Urban non-teaching	361 (22.4)	363 (23.7)		277 (23.8)	272 (23.4)
Urban teaching	1152 (71.6)	1089 (71.2)		816 (70.2)	826 (71.1)
Hospital Region, n (%)			0.084			0.874
Northeast	325 (20.2)	301 (19.7)		233 (20.1)	235 (20.2)
Midwest	388 (24.1)	430 (28.1)		296 (25.5)	312 (26.9)
South	525 (32.6)	467 (30.5)		370 (31.8)	361 (31.1)
West	371 (23.1)	332 (21.7)		263 (22.6)	254 (21.9)
Hospital Bed Size, n (%)			0.628			0.738
Small	235 (14.6)	209 (13.7)		169 (14.5)	156 (13.4)
Medium	439 (27.3)	408 (26.7)		310 (26.7)	315 (27.1)
Large	935 (58.1)	913 (59.7)		683 (58.8)	691 (59.5)
Comorbidities, n (%)
Smoking	701 (43.6)	669 (43.7)	0.958	506 (43.5)	507 (43.6)	1
Hypertension	656 (40.8)	881 (57.6)	<0.001	600 (51.6)	587 (50.5)	0.619
DM	203 (12.6)	421 (27.5)	<0.001	197 (17.0)	225 (19.4)	0.146
Obesity	162 (10.1)	236 (15.4)	<0.001	133 (11.4)	162 (13.9)	0.081
Anxiety	397 (24.7)	396 (25.9)	0.461	292 (25.1)	301 (25.9)	0.703
Depression	247 (15.4)	307 (20.1)	0.001	203 (17.5)	222 (19.1)	0.334
COPD	353 (21.9)	331 (21.6)	0.87	265 (22.8)	260 (22.4)	0.843
OSA	62 (3.9)	109 (7.1)	<0.001	56 (4.8)	67 (5.8)	0.354
CKD	110 (6.8)	164 (10.7)	<0.001	104 (9.0)	121 (10.4)	0.262
PAD	80 (5.0)	134 (8.8)	<0.001	70 (6.0)	76 (6.5)	0.669
Sepsis	30 (1.9)	17 (1.1)	0.112	15 (1.3)	16 (1.4)	1

Abbreviations: CKD, chronic kidney disease; COPD, chronic obstructive pulmonary disease; DM, diabetes mellitus; HLD, hyperlipidemia; OSA, obstructive sleep apnea; PAD, peripheral artery disease; TCM, takotsubo cardiomyopathy.

Figure 2

Standardized mean differences of covariates before and after propensity score matching between TCM patients with and without HLD.

In-Hospital Mortality, Length of Stay, and Total Charges

Prior to propensity score matching, there was no statistical difference in in-hospital mortality between the two groups. We found that HLD patients had a shorter LOS (3.26 ± 3.17 days vs 3.60 ± 3.46 days, p = 0.004) and lower total charges of stay (US$ 47,048.09 ± 40,379.55 vs US$ 52,478.42 ± 54,067.55, p = 0.002) compared to TCM patients without HLD (Table 3).

Table 3

In-Hospital Outcomes Before and After Propensity Score Matching

Variables	Unmatched Cohort			Propensity-Matched Cohort
	TCM without HLD	TCM with HLD	P value	TCM without HLD	TCM with HLD	P value
n	1609	1530		1162	1162
Outcomes
Death, n (%)	33 (2.1)	19 (1.2)	0.102	28 (2.4)	13 (1.1)	0.027
Cardiac arrest, n (%)	37 (2.3)	19 (1.2)	0.035	24 (2.1)	14 (1.2)	0.141
Cardiogenic shock, n (%)	91 (5.7)	54 (3.5)	0.006	65 (5.6)	39 (3.4)	0.012
Ventricular arrhythmia	70 (4.4)	53 (3.5)	0.235	46 (4.0)	41 (3.5)	0.662
AKI, n (%)	146 (9.1)	141 (9.2)	0.940	114 (9.8)	94 (8.1)	0.167
ARF, n (%)	200 (12.4)	142 (9.3)	0.006	141 (12.1)	106 (9.1)	0.022
LOS, (mean (sd))	3.60 (3.46)	3.26 (3.17)	0.004	3.57 (3.14)	3.20 (3.27)	0.005
Total charge (mean (sd))	52,478.42 (54,067.55)	47,048.09 (40,379.55)	0.002	51,282.36 (46,190.15)	46,825.87 (39,620.81)	0.013

Abbreviations: AKI, acute kidney injury; ARF, acute respiratory failure; HLD, hyperlipidemia; LOS, length of stay; TCM, takotsubo cardiomyopathy.

After propensity score matching, we found that the in-hospital mortality rate was lower in HLD group (1.1% vs 2.4%, p=0.027; OR 0.46, 95% CI 0.24–0.89). (Figure 3) HLD group has shorter LOS (3.20 ± 3.27 days vs 3.57 ± 3.14 days, p = 0.005) and less total charges of stay (US$ 46,825.87 ± 39,620.81 vs US$ 51,282.36 ± 46,190.15, p = 0.013) (Table 3).

Figure 3

Forrest plot graph showing adjusted odds ratio for in-hospital outcomes after propensity score matching.

In-Hospital Complications

We found that the HLD group had a lower incidence of in-hospital complications, including cardiac arrest (1.2% vs 2.3%, p = 0.035), cardiogenic shock (3.5% vs 5.7%, p = 0.006) and ARF (9.3% vs 12.4%, p = 0.006) in unmatched cohorts. There was no statistical significance when comparing the outcomes of ventricular arrhythmias and AKI between the two groups.

After propensity score matching, the HLD group had lower incidence of cardiogenic shock (3.4% vs 5.6%, p = 0.012; OR 0.59, 95% CI 0.39–0.88) and ARF (9.1% vs 12.1%, p = 0.022; OR 0.73, 95% CI 0.56–0.95). The details of the result can be found in Table 3 and Figure 3.

Discussion

To our knowledge, this is the first study to examine the association between HLD and the short-term outcomes of patients being admitted to hospital for TCM. We found that patients who carried a diagnosis of HLD and were admitted for TCM were associated with lower in-hospital mortality rate, shorter LOS, lower total charges of stay and lower rate of cardiogenic shock.

Similar to previous studies, we found that HLD was common among TCM patients, with 48.7% of them carrying a diagnosis of HLD.12–14 Though the pathophysiology of TCM is not well understood, endothelial dysfunction is one of the well-accepted hypotheses.29,30 HLD is known to cause endothelial dysfunction as a result of an elevated oxidized LDL-C (ox-LDL) level. Ox-LDL can damage vascular endothelium by 1) downregulating Hippo Yes-associated protein/YLP motif contain 1 (Hippo-YAP/ZAP) pathway which in turn interferes with cardiovascular remodeling,31 2) by inducing the apoptosis of vascular endothelial as a result of upregulated expression of autophagy-related protein,32 and 3) by inhibiting vascular relaxations induced by nitric oxide.33 In theory, HLD patients are more likely to have endothelial dysfunction, which may facilitate the pathophysiologic basis for the occurrence of TCM.

Interestingly, our study found TCM patients with HLD had paradoxically better in-hospital outcomes than those without HLD – which parallels the previous findings in other cardiovascular diseases such as acute myocardial infarction34–36 and heart failure.37,38 Studies have suggested that patients with myocardial infarction with higher cholesterol and TG levels were associated with better short-term and long-term outcomes.34–36 A lower level of serum cholesterol was found to be associated with the worse one-year outcomes in heart failure patients,37,38 and the mortality difference between the low cholesterol group and high cholesterol group can be appreciated after 30-day follow up.39 Likewise, type-A acute aortic dissection patients with lower cholesterol levels have a higher in-hospital mortality as well.40

Postulated mechanisms of this phenomenon are as follows. First, sepsis, as the third leading cause of short-term41 and long-term42 cause of death in TCM, is accountable for 21.6% of inpatient mortality among patients with TCM.41 Lipoproteins may downregulate the inflammatory immune response via interaction between lipoproteins and bacterial lipopolysaccharide (LPS) in sepsis, and LDL-C has been shown to bind LPS and to protect against the immediate toxic effects of LPS on endothelial cells.43 The higher level of lipoproteins in HLD patients may have a protective effect for sepsis which is a major cause of death in TCM.

In addition, with the use of statin as the first-line treatment of HLD, patients in HLD group most likely benefit from the statins’ anti-inflammation effect.44 For example, atorvastatin, one of the most widely used statins, was found to decrease the expression of inflammatory factors, such as interleukin-6 (IL-6),45 which is associated with higher incidence of adverse event from TCM.46 Another retrospective study of 1617 myocardial infarction patients also found that long-term statins use was associated with lower C-reactive protein levels and better myocardial reperfusion. This anti-inflammation benefit was independent of its lipid-lowering effect.47 Though a retrospective study by Dias et al involving 146 participants who received statin during hospitalization found that statins was not associated with major adverse cardiovascular events in TCM patients after multivariate analysis,14 only 96 of them had a history of HLD, and the statins were prescribed in an inpatient setting, which may not have been reflective of the long-term benefit of the statins. Patients with an established diagnosis of HLD are more likely to be on a statin chronically and may be able to benefit most from the anti-inflammatory effects, though more prospective study is warranted to verify this hypothesis.

Finally, compared to patients with HLD, those without HLD may include a subset of individuals with low cholesterol and TG levels. A previous study found that cholesterol level had a strong correlation with pre-albumin – a strong indicator of nutritional status.48 Data from National Health and Nutrition Examination Surveys has demonstrated that TG levels had a significant correlation with Body Mass Index,49 which is another common indicator of nutritional status. Therefore, we inferred that part of patients in non-HLD group to have low nutrition status in theory, though we could not obtain data on nutritional status (such as LDL-C, pre-albumin and weight) from the NIS database. Low cholesterol and TG level may reflect the state of malnutrition and cachexia - this may be one of the factors that resulted in worse outcomes for TCM patients without HLD.

The study has several strengths. First, given the rarity of TCM, we used the latest NIS from both 2016 and 2017 – the largest national inpatient database and identified 3139 patients. Second, we created an adjusted cohort with a balanced baseline characteristics and comorbidities by using propensity score matching. By doing so, we were able to minimize the confounding factors to determine a true correlation. Lastly, this is the first study to determine the correlation between HLD and the outcomes of patients admitted in hospital for TCM.

There are several limitations in our study. First, due to a retrospective nature of the study, we can only minimize confounding factors, not eliminate the confounding factors. In addition, the comorbidities were selected based on ICD-10-CM codes, and the severity and duration of comorbidities could not be ascertained. Moreover, given the nature of NIS data, certain data on lipid levels (such as serum cholesterol, triglyceride level) and nutritional status (such as weight and pre-albumin levels) were not available, which prevents us from analyzing those factors. Similarly, treatment information (such as statin, non-steroidal anti-inflammatory drugs, steroids and anti-inflammatory agents) were not included in the NIS data. These details are important to determine the impact of lipid and lipid-lowering agents on TCM. Moreover, certain mechanical features of TCM, such as typical versus atypical, may play a role in the outcomes, but were again not available in the NIS.

Conclusion

We conclude that HLD is paradoxically associated with better in-hospital outcomes of TCM hospitalization. TCM patients with HLD have a lower in-hospital mortality rate, lower incidence of cardiogenic shock and ARF, shorter LOS and lower total charges of stay. Further studies are needed to investigate the factors and medical therapies that may contribute to reductions of in-hospital mortality in TCM patients with HLD. Moreover, a more complete understanding of the role of lipid metabolism in TCM patients may help guide treatment strategies and maximize the benefit of the effect of dyslipidemia treatment.