Congestive Heart Failure

In-hospital mortality among patients with takotsubocardiomyopathy: A study of the National InpatientSample 2008 to 2009Waleed Brinjikji, MD, a Abdulrahman M. El-Sayed, DPhil, b,c and Samer Salka, MD, FACC d Dearborn, MI;and New York, NY

Background Takotsubo cardiomyopathy is characterized by acute, reversible left ventricular apical ballooning. Little isknown about the characteristics of patients with takotsubo cardiomyopathy who have in-hospital mortality. We sought todetermine in-hospital mortality rate, complication rate, and characteristics of patients with in-hospital mortality related totakotsubo cardiomyopathy.

Methods Patients diagnosed with takotsubo cardiomyopathy in the National Inpatient Database Samples 2008 to 2009using International Classification of Diseases, Ninth Revision, code 42983 were included in this study. Our primary outcomewas in-hospital mortality. In patients with takotsubo cardiomyopathy, we assessed demographic factors, the prevalence andassociated mortality of underlying critical illnesses (acute ischemic stroke, sepsis, acute renal failure, respiratory insufficiency,and noncardiac surgery), and acute complications (acute congestive heart failure, respiratory insufficiency with congestiveheart failure, cardiogenic shock, ventricular fibrillation/cardiac arrest, and intraaortic balloon pump placement).

Results A total of 24,701 patients with takotsubo cardiomyopathy were identified. In-hospital mortality rate was 4.2%. Atotal of 21,994 patients (89.0%) were female. Male patients had a higher mortality rate than females (8.4% vs 3.6%, P b

.0001). Age and race were not associated with mortality. Of patients with in-hospital mortality, 81.4% had underlying criticalillnesses. Male patients with takotsubo had higher incidence of underlying critical illnesses than their female counterparts(36.6% vs 26.8%, P b .0001).

Conclusions The presence of underlying critical illness was the main driver of mortality, as these patients comprisedN80% of patients with in-hospital mortality. Male patients, who were significantly more likely to have underlying critical illness,had significantly higher mortality rates than female patients. The presence of underlying critical illness likely explains the highermortality rate among male patients. (Am Heart J 2012;164:215-21.)

Takotsubo cardiomyopathy is characterized by acute,reversible, left ventricular (LV) dysfunction with wallmotion abnormalities of the middistal and apical regionsof the LV.1-3 Patients with takotsubo cardiomyopathy aregenerally able to fully recover myocardial function withina period of days to weeks.4 Little is known regardingpredictors of poor outcomes in patients with takotsubocardiomyopathy. In this analysis, we used the NationalInpatient Sample (NIS) between 2008 and 2009 to

From the aDepartment of Medical Education, Oakwood Hospital, Dearborn, MI, bCollegeof Physicians and Surgeons, Columbia University, New York, NY, cDepartment ofEpidemiology, Columbia University, New York, NY, dPremier Cardiovascular Specialists,Dearborn, MI.Submitted January 19, 2012; accepted April 22, 2012.Reprint requests: Samer Salka, MD, Premier Cardiovascular Specialists, 15120 MichiganAve, Dearborn, MI 48126.E-mail: brinjikji.waleed@gmail.com0002-8703/$ - see front matter© 2012, Mosby, Inc. All rights reserved.doi:10.1016/j.ahj.2012.04.010

determine the prevalence of concomitant severe illness,complications, and mortality associated with takotsubocardiomyopathy as well as to determine the demographicand clinical predictors of mortality among patients withtakotsubo in the United States.

MethodsPatient selectionThis study was deemed institutional review board exempt by

our institution as the NIS is a publically available administrativedatabase. We purchased the NIS hospital discharge databasefor the period 2008 to 2009 from the Healthcare Cost andUtilization Project of the Agency for Healthcare Research andQuality, Rockville, MD. The NIS is a hospital dischargedatabase representing 20% of all inpatient admissions tononfederal hospitals in the United States. We included allpatients with a diagnosis of takotsubo cardiomyopathy (code42983 in the International Classification of Diseases, NinthRevision, Clinical Modification).

Table I. Demographic characteristics and mortality among allpatients with takotsubo cardiomyopathy from the NIS, 2008 to2009

Takotsubopatients

n (%),mortality

Unadjustedmortality

OR (95% CI)

n 24701 1027 (4.2) –Age, mean ± SD 66.9 ± 30.7 – –Age group

b50 y 2689 (10.9) 105 (3.9) Ref50-64 y 7290 (29.5) 245 (3.4) 0.86 (0.68-1.08)N64 y 14722 (59.6) 677 (4.6) 1.19 (0.96-1.46)

GenderFemale, n (%) 21994 (89.0) 799 (3.6) RefMale, n (%) 2707 (11.0) 228 (8.4) 2.44 (2.09-2.84)⁎

Race, n (%)White 16680 (84.0) 668 (4.0) RefBlack 1178 (5.9) 49 (4.2) 1.04 (0.77-1.40)Hispanic 1032 (5.2) 50 (4.9) 1.22 (0.91-1.64)Asian 353 (1.8) 15 (4.2) 1.06 (0.63-1.79)

Mean ± SD CCI 1.4 ± 2.7 – –Chronic comorbidities

Obesity 1494 (6.1) 29 (2.0) 0.44 (0.31-0.64)⁎HTN 14434 (58.4) 428 (3.0) 0.49 (0.44-0.56)⁎Hyperlipidemia 9261 (37.5) 119 (1.3) 0.21 (0.17-0.25)⁎Diabetes mellitus 4661 (18.9) 157 (3.4) 0.77 (0.64-0.91)⁎Smoking 3250 (13.2) 81 (2.5) 0.56 (0.44-0.70)⁎Malignancy 3547 (14.4) 288 (8.1) 2.45 (2.13-2.82)⁎Anxiety disorder 2204 (8.9) 22 (1.0) 0.22 (0.14-0.34)⁎Mood disorder 3696 (15.0) 67 (1.8) 0.39 (0.30-0.50)⁎

HTN, Hypertension; Ref, reference.⁎ P b .001.

216 Brinjikji et alAmerican Heart Journal

August 2012

Demographics and patient factorsDemographic factors studied included age, gender, and race.

We determined mean age and SD and stratified patients into 3arbitrarily selected age groups: (1) patients b50 years, (2)patients 50 years old to 64 years old, and (3) patients ≥65 years.Races studied included white, black, Hispanic, and Asian/PacificIslander. Charlson comorbidity index (CCI) was calculated foreach patient.5 We also report the prevalence and associatedmortality of the following chronic comorbidities: diabetesmellitus, hypertension, hyperlipidemia, smoking, obesity,mood disorders, anxiety disorders, and malignancy.

OutcomesThe primary outcome of our studywas in-hospitalmortality.We

determined the prevalence and associatedmortality of underlyingcritical illness as well as complications/procedures related totakotsubo cardiomyopathy. Complications/procedures studiedincluded cardiogenic shock, ventricular fibrillation (Vfib)/cardiacarrest, acute congestive heart failure (CHF), intraaortic balloonpump (IABP) placement, and respiratory insufficiency with acuteCHF. Underlying critical illnesses considered included acute renalfailure, sepsis, acute cerebrovascular accident (CVA), andrespiratory insufficiency without acute CHF. We used endotra-cheal tube placement/mechanical ventilation as a marker forrespiratory distress. We assumed that patients who wereintubated and/or mechanically ventilated without a diagnosis ofacute CHF were likely intubated before having takotsubocardiomyopathy, whereas those who were intubated/mechani-cally ventilatedwith CHFweremore likely to have undergone thisprocedure as a result of CHF associated with takotsubocardiomyopathy. Mean length of stay was also calculated.

Statistical analysisWe summed data from 2008 to 2009. Categorical variables are

presented as n (%), and continuous variables are presented asmean ± SD. We determined the association between each of theaforementioned categorical demographic and patient factorswith each of the aforementioned complications and in-hospitalmortality using χ2 tests. Odds ratios (ORs) with 95% CIs (inparentheses) and associated P values are presented for each ofthese analyses. To determine which demographic and patientfactors were most predictive of in-hospital mortality after mutualadjustment, we performed a multivariable logistic regressionmodel adjusting for patient age, CCI, gender, race, and presenceof underlying critical illness. Critical illness was used as adichotomous variable. Statistical analyses were performed usingthe SAS-based statistical package JMP 9.0 (www.jmp.com) andSAS 9.1 (www.sas.com).

No extramural funding was used to support this work. Theauthors are solely responsible for the design and conduct of thisstudy, all study analyses, the drafting and editing of themanuscript, and its final contents.

ResultsDemographic characteristicsAmong 24,701 patients, a large majority (21,994, 89%)

were female. Mean age was 66.9 ± 30.7 years, and most

patients (59.6%) were ≥65 years old. Male patients weregenerally younger than their female counterparts (62.4 ±37.3 years vs 67.5 ± 29.6 years, respectively, P b .0001).Mean CCI was 1.4 ± 2.7 with no difference between maleand female patients (1.5 ± 3.1 vs 1.4 ± 2.6, respectively,P = .28). Eighty-four percent of patients with takotsubowere white. The most prevalent chronic comorbiditieswere hypertension (58.4%), hyperlipidemia (37.5%), anddiabetes mellitus (18.9%). These data are summarized inTable I.

OutcomesAmong patients admitted to the hospital with diagnosis

of takotsubo cardiomyopathy, 1,027 (4.2%) had in-hospital mortality. Mean length of stay was 6.1 ± 15.9days. Neither age nor race was associated with mortality.Male gender was associated with higher mortality odds(OR 2.44, 95% CI 2.09-2.84). The only chronic comor-bidity associated with a higher odds of in-hospitalmortality was malignancy (OR 2.45, 95% CI 2.13-2.82).Prevalence of underlying critical illness, complications,

and procedures are presented in Table II, and mortalityrates associatedwith eachof these covariates are presentedin Table III. The total prevalence of acute complications(acute CHF, IABP placement, cardiogenic shock, Vfib/

Table II. Acute complications and underlying illness overall and by gender among all patients with takotsubo cardiomyopathy from the NIS,2008 to 2009

All takotsubo Male FemaleOR (95% CI),male/female

Acute complicationsCardiogenic shock 1162 (4.7) 155 (5.7) 1006 (4.6) 1.27 (1.07-1.51)⁎Vfib/cardiac arrest 686 (2.8) 111 (4.1) 575 (2.6) 1.59 (1.29-1.96)†

Respiratory insufficiency with acute CHF 1661 (6.7) 202 (7.4) 1459 (6.6) 1.13 (0.97-1.32)Acute CHF 7551 (31.1) 712 (26.3) 6839 (31.1) 0.79 (0.72-0.87)†

IABP 766 (3.1) 88 (3.2) 678 (3.1) 1.05 (0.84-1.31)Any acute complication 8640 (34.5) 901 (33.3) 7739 (35.2) 0.92 (0.84-1.00)

Underlying critical illnessAcute renal failure 2586 (10.5) 426 (15.7) 2160 (9.8) 1.71 (1.53-1.92)†

Sepsis 1761 (7.1) 336 (12.4) 1426 (6.5) 2.04 (1.80-2.32)†

Respiratory insufficiency without Acute CHF 2084 (8.4) 373 (13.8) 1711 (7.8) 1.58 (1.31-1.90)†

Stroke/TIA 655 (2.7) 93 (3.5) 562 (2.6) 1.36 (1.09-1.71)⁎SAH 223 (0.9) 14 (0.5) 209 (1.00) 0.55 (0.32-0.95)⁎Noncardiac OR procedures 2517 (10.1) 292 (10.8) 2225 (10.1) 1.07 (0.94-1.22)Any underlying critical illness 6892 (27.9) 991 (36.6) 5901 (26.8) 1.57 (1.45-1.71)†

TIA, Transient ischemic attack; SAH, subarachnoid hemorrhage.⁎ P b .05.† P b .001.

Table III. Association between acute complications and underlying illness and NIS, 2008 to 2009

n (%),died

OR (95% CI),mortality

Male,n (%), died

OR (95% CI)mortality, males

Female,n (%), died

OR (95% CI)mortality, females

Acute complicationsCardiogenic shock 186 (16.0) 5.15 (4.32-6.11)⁎ 41 (26.3) 4.50 (3.05-6.62)⁎ 145 (14.4) 5.25 (4.33-6.37)⁎Vfib/cardiac arrest 187 (27.3) 10.4 (8.66-12.45)⁎ 37 (33.7) 6.41 (4.21-9.75)⁎ 150 (26.1) 11.31 (9.24-13.84)⁎Respiratory insufficiencywith CHF

254 (15.3) 5.22 (4.48-6.07)⁎ 54 (26.6) 4.86 (3.43-6.88)⁎ 201 (13.8) 5.32 (4.49-6.30)⁎

Acute CHF 381 (5.0) 1.36 (1.19-1.54)⁎ 74 (10.4) 1.40 (1.04-1.87)⁎ 306 (4.5) 1.40 (1.21-1.61)⁎IABP 142 (18.5) 5.91 (2.87-7.19)⁎ 20 (23.2) 3.52 (2.10-5.89)⁎ 121 (17.9) 6.64 (5.38-8.20)⁎Any acute complication 610 (7.1) 2.85 (2.51-3.24)⁎ 134 (14.9) 3.18 (2.41-4.20)⁎ 476 (6.2) 2.83 (2.45-3.27)⁎

Underlying critical illnessAcute renal failure 358 (13.9) 5.16 (4.51-5.92)⁎ 114 (26.8) 6.95 (5.22-9.24)⁎ 244 (11.3) 4.44 (3.7905.19)⁎Sepsis 380 (21.6) 9.48 (8.26-10.89)⁎ 84 (24.9) 5.12 (3.80-6.91)⁎ 296 (20.8) 10.48 (8.97-12.25)⁎Respiratory insufficiencywithout CHF

491 (23.5) 12.66 (11.09-14.46)⁎ 135 (36.1) 13.57 (10.10-18.24)⁎ 356 (20.8) 11.75 (10.11-13.65)⁎

Stroke/TIA 123 (18.7) 5.90 (4.80-7.26)⁎ 24 (26.2) 4.20 (2.59-6.80)⁎ 98 (17.5) 6.28 (4.99-7.91)⁎SAH 79 (35.2) 13.51 (10.18-17.91)⁎ 11 (78.6) 23.0 (7.56-70.07)⁎ 69 (33.1) 14.26 (10.59-19.21)⁎Noncardiac OR procedures 202 (8.0) 2.25 (1.92-2.64)⁎ 25 (8.7) 1.03 (0.67-1.59) 176 (7.9) 2.65 (2.23-3.15)⁎Any underlying critical illness 836 (12.1) 12.72 (10.84-14.93)⁎ 197 (19.9) 13.23 (9.06-19.61)⁎ 639 (10.8) 12.11 (10.15-14.44)⁎

⁎ P b .0001.

Brinjikji et al 217American Heart JournalVolume 164, Number 2

cardiac arrest, respiratory insufficiency with acute CHF)associated with takotsubo cardiomyopathy was 34.5%(33.3% inmales and 35.2% in females, P = .051). The rate ofmortality among patients with acute complications was7.1% (14.9% inmales and 6.2% in females,P b .0001). Maleshad higher prevalence of cardiogenic shock and Vfib,whereas females had higher prevalence of acute CHF. Themost common acute complication was acute CHF, presentin 31.1% of the population, and associated with a 5.0%mortality rate. Respiratory insufficiency in the setting ofCHF was present in 6.7% of the population and had an

associatedmortality rate of 15.3%. Cardiogenic shock, Vfib,and IABP placement were each present in b5% of thepopulation but were associated with significantly higherodds of mortality.The prevalence of underlying critical illness among

patients with takotsubo was 27.9% (6,892 patients). Ofthe male and female patients, 36.6% and 26.8%,respectively, had underlying critical illness (P b .0001).Mortality rate among these patients was 12.1% with amortality rate of 19.9% in males and 10.8% in females(P b .0001). Of the 1,027 patients who had in-hospital

Table IV. Multivariate analysis results

OR (95% CI) P

Age groupb50 y Ref Ref50-64 y 1.01 (0.77-1.32) .95N64 y 1.04 (0.82-1.35) .73

GenderFemale Ref RefMale 2.07 (1.71-2.49) b.0001

RaceWhite Ref RefBlack 0.87 (0.63-1.17) .35Hispanic 0.92 (0.67-1.24) .59Asian 0.65 (0.36-1.09) .10

CCI⁎ 1.19 (1.13-1.26) b.0001Underlying critical illness

No Ref RefYes 10.87 (9.08-13.08) b.0001

⁎Odds ratio for each unit increase in CCI.

218 Brinjikji et alAmerican Heart Journal

August 2012

mortality, 836 patients had underlying critical illnessduring their hospitalization (81.4%). Of patients who didnot have any critical illness (n = 17,809), 191 had in-hospital mortality (1.1%). All underlying critical illnesseswere associated with higher risk of mortality, werepresent more often in males, and were associated withhigher mortality rates for male patients (P b .0001 forall comorbidities). Cerebrovascular accident was presentin 2.7% of patients with takotsubo cardiomyopathy.Sepsis, present in 7.1% of the population, was associatedwith a 21.6% mortality rate among patients withtakotsubo, whereas respiratory insufficiency withoutacute CHF, present in 8.4% of patients, was associatedwith a 23.5% mortality rate. A total of 2,516 patients withtakotsubo (10.1%) underwent noncardiac OR proceduresduring their hospitalization with a mortality rate of 8.0%.These data are summarized in Tables II and III.In our multivariate analysis, age and race were not

independently associated with increased mortality. Malegender was independently associated with increasedmortality (OR 2.07, 95% CI 1.71-2.49) as was CCI (OR1.19, 95% CI 1.13-1.26). The presence of critical illnesswas the strongest independent variable associated withincreased mortality (OR 10.87, 95% CI 9.08-13.08). Thesedata are summarized in Table IV.

DiscussionOur study of the in-hospital mortality of nearly 25,000

patients with takotsubo cardiomyopathy allowed for anumber of important observations. Overall, the nearly30% of patients who had underlying critical illnesses orunderwent noncardiac surgeries during their hospital-ization comprised nearly 80% of patients who had in-hospital mortality. The in-hospital mortality rate forpatients with takotsubo cardiomyopathy was 4.2% with

higher mortality rates for males than females (8.4% vs3.6%). Males were significantly more likely to haveunderlying critical illnesses associated with takotsubocardiomyopathy than females. Furthermore, males, ingeneral, had higher rates of acute complications(except acute CHF) and in-hospital mortality associatedwith these complications than their female counter-parts. Of the chronic comorbidities studied, onlymalignancy was associated with a higher odds of in-hospital mortality.The overall in-hospital mortality rate of 4.2% is

consistent with previous studies, which demonstratemortality rates ranging from 0% to 8%.6-11 This in-hospitalmortality rate is similar to those of myocardial infarc-tion.12 In our study, we assessed demographic andcomorbid factors that were predictive of mortality; wedemonstrated that male gender, increased CCI, andpresence of underlying critical illness were independent-ly associated with increased mortality after adjusting forage, race, gender, CCI, and presence of underlyingcritical illness. Previous studies have not demonstratedcomparative mortality rates between male and femalepatients diagnosed with takotsubo cardiomyopathy,although this is likely a result of low power in previousstudies, a product of the small sample sizes of thesestudies, low mortality rate of takotsubo cardiomyopathy,and the low number of males diagnosed with thiscondition. Race and increased age were not associatedwith in-hospital mortality.The most common complication in our study was acute

CHF (31.1%) with a mortality rate of 5.0%. In a previousstudy examining the incidence and severity of heartfailure among patients diagnosed with takotsubo cardio-myopathy, Madhavan et al9 demonstrated the incidenceof acute CHF was 45% with 3% of patients havingin-hospital mortality secondary to cardiogenic shock. Fivepercent of patients in our study had cardiogenic shock,and 3% had Vfib or cardiac arrest. Sharkey et al13

demonstrated in their case series that the incidence ofcardiac arrest among hospitalized patients with takotsubowas approximately 5%. In a systematic review, Gianniet al4 demonstrated the incidence of cardiogenic shockand Vfib to be 4.2% and 1.5%, respectively. The use ofIABP has been well described for patients with takotsubocardiomyopathy with cardiogenic shock. Eitel et al7

reported implantation of IABP in 3% of patients overall.Intraaortic balloon pump has been reported in 0% to 18%of cases depending on the case series.6-11 In our study,we demonstrated the use of IABP in approximately 3%of cases. As expected, patients requiring this treatmenthad high mortality rates (18.5%).Respiratory insufficiency necessitating mechanical ven-

tilation and endotracheal tube placement was relativelycommon in our sample (15.2%). The incidence ofmechanical ventilation for patients with takotsubo hasbeen reported as 12% to 15%.9,13,14 Often, this is

Brinjikji et al 219American Heart JournalVolume 164, Number 2

associated with acute CHF.9 Respiratory insufficiency mayboth be a precipitant of or a complication of takotsubo. Itis plausible that patients receiving mechanical ventilationin the setting of takotsubo and acute CHF likely requiredtreatment secondary to takotsubo, whereas patientsrequiring ventilation without acute CHF likely hadtakotsubo as a result of critical illness, although thetemporal relationship is impossible to ascertain from thedata. In our study, 34.1% of patients with respiratoryinsufficiency had acute CHF, and the associated mortalitywas lower than in respiratory insufficiency patientswithout acute CHF (15.3% vs 23.5%, respectively). Patientswho had in-hospital mortality and underwent mechanicalventilation with a concomitant diagnosis of CHF com-prised approximately 1% of the total study population.Approximately 30% of the patients in our study had

associated underlying noncardiac surgery, acute renalfailure, sepsis, CVA, or respiratory insufficiency. Criticalillness, most notably sepsis, has been described as aprecipitant of takotsubo cardiomyopathy in previousstudies and case reports.14-18 In a series by Lee et al,14

sepsis and hypoxemia were considered the triggeringevent for takotsubo cardiomyopathy in approximately52% of patients. Sepsis was not predictive of mortality;however, this is likely due to the low power of this studyto detect factors associated with mortality. Acute renalfailure has not been described in the literature as aconsequence of takotsubo cardiomyopathy, and thus, weassumed that this condition preceded diagnosis as well.However, the combination of LV dysfunction resultingfrom takotsubo cardiomyopathy in the setting of criticalillnesses such as hypoxia and sepsis may explain the highrate of acute renal failure.Takotsubo in the postoperative setting has been

described in previous series and case reports.14,19 Leeet al14 demonstrated that approximately 30% of patientsin their series of patients with takotsubo had an in-hospital surgery or procedure. Our study demonstratedthat approximately 10% of patients underwent a non-cardiac procedure during their hospitalization and thatthis was associated with a higher mortality rate in general.The mechanism behind this is unclear; however, it ispossible that elevated circulating catecholamines in thepostoperative state are a contributing factor.Acute CVA associated with takotsubo cardiomyopathy

had a mortality rate of 18%. Cerebrovascular accident hasbeen described in the literature as both a precipitant20

and result of takotsubo.11 In a review of 569 consecutivepatients with acute ischemic stroke admitted to a strokecenter, Yoshimura et al11 reported that 7 patients werediagnosed as having takotsubo cardiomyopathy. Abnor-mal findings on electrocardiography occurred within 10hours of stroke onset in 5 patients and within 2 weeks in2 patients. The frequency of acute ischemic strokeprecipitating takotsubo cardiomyopathy has not beenestablished in previous case series. Takotsubo cardiomy-

opathy resulting in CVA is thought to be caused by muralthrombus formation secondary to the wall motionabnormalities characteristic of this condition.20

Overall, the nearly 30% of patients who had underlyingcritical illnesses comprised N80% of patients who hadin-hospital mortality. Among patients who did not havethese critical illnesses, the mortality rate was only 1.1%.These data suggest that patients who are diagnosed withtakotsubo cardiomyopathy in the setting of underlyingcritical illness have substantially higher risk of in-hospitalmortality and make up the large majority of patientswith takotsubo with in-hospital mortality. Conversely,uncomplicated takotsubo cases are likely to be dis-charged alive in nearly 99% of cases. These findingssuggest that the main driver of in-hospital mortality inpatients with takotsubo cardiomyopathy is the pre-sence of underlying critical illness. These findings arecorroborated by a study published by Song et al,21 whichdemonstrated no cases of cardiac mortality amongpatients diagnosed with takotsubo cardiomyopathy.The study, which demonstrated an in-hospital mortalityrate of 9%, found that all patients with in-hospitalmortality died of noncardiac causes. In our study, weare unable to elucidate whether mortality was secondaryto cardiac or noncardiac causes; however, the fact thatmost patients with in-hospital mortality had underlyingcritical illness suggests that most mortality was due tononcardiac causes.In general, males had significantly higher prevalence of

all critical illnesses. Male patients were N50% more likelyto be diagnosed with an underlying critical illness, andfurthermore, those males with concomitant underlyingcritical illness comprised N85% of in-hospital male deaths.However, on our multivariate analysis, when controllingfor the presence of underlying critical illness, malegender remained significantly associated with increasedin-hospital mortality. The reason behind the associationof increased mortality among males independent of theirhigher rate of underlying critical illness is unclear. Malepatients also had higher rates of complications such asVfib/cardiac arrest, cardiogenic shock, and respiratoryinsufficiency. We believe that the higher rate ofunderlying critical illness in this population is likely amajor contributor to this finding.

LimitationsThis study has several limitations. First, we were unable

to assess long-term morbidity and mortality, as the NISdatabase does not provide any follow-up data on patients.However, it is unlikely that longer follow-up would havechanged our findings appreciably, as the literaturesuggests that a large majority of mortality occurs in thefirst few days after onset of LV dysfunction.14 Recurrenceof takotsubo cardiomyopathy is a very rare event;however, some studies have suggested that the rate of

220 Brinjikji et alAmerican Heart Journal

August 2012

cardiac rehospitalization is not insignificant.22,23 Second,coding errors are a potential limitation of this study, asthey are of any study from an administrative database.However, in general, our findings are corroborated bythose of previously published case series, giving us littlereason to believe that any miscoding of takotsubocardiomyopathy in our data set significantly altered ourfindings or introduced any significant bias. No previousstudies have assessed the accuracy of the takotsubocardiomyopathy International Classification of Dis-eases, Ninth Revision, code in administrative databases.We assume that acute comorbidities were present beforeonset of takotsubo cardiomyopathy, whereas acutecomplications resulted from takotsubo cardiomyopathy.Unfortunately, the NIS database does not provide atimeline for diagnoses, and thus, we cannot be certainthat these assumptions are appropriate.Given our use of the NIS database, we were unable to

assess the frequency and prognosis of a number ofimportant previously published prognostic factors.Given the fact that a number of variants of takotsubocardiomyopathy exist but only 1 International Classi-fication of Diseases, Ninth Revision, code exists for thesyndrome, we cannot determine the frequency andoutcomes of certain variants, such as the midventricularvariant in our study.24 The lack of our ability tocorrelate imaging findings with outcomes in our studyis also a major limitation. A number of echocardiograph-ic findings have been associated with poor outcomes,especially right ventricular dysfunction. Right ventricu-lar involvement in takotsubo cardiomyopathy is associ-ated with more severe impairment of LV systolicfunction14,25 and longer hospital stays. The associationbetween echocardiographic findings and mortality hasnot been established.26

In addition to the above limitations, misdiagnosis oftakotsubo is a potential concern. Our study demonstratedthat takotsubo cardiomyopathy is associated with criticalillness in approximately 30% of cases. Critical illnesses,especially sepsis, are associated with a number ofcardiomyopathies that could be confused with takotsubo.Sepsis-associated myocardial dysfunction, similar totakotsubo cardiomyopathy, is characterized by a revers-ibly depressed LV ejection fraction and mild increases incardiac biomarkers.27 If patients can recover from theinitial insult, they often demonstrate a complete recoveryof LV function. Importantly, however, this condition isnot associated with characteristic imaging findings astakotsubo cardiomyopathy is. The relationship anddistinguishing features between these 2 diseases havenot been well established. Although it is entirely possiblethat patients with sepsis-associated myocardial dysfunc-tion were classified as having takotsubo cardiomyopathy,given the strict imaging criteria needed for a diagnosis oftakotsubo cardiomyopathy, we believe that it is unlikelythat this was a systematic error.28

ConclusionsIn this retrospective study of a large administrative

database, we demonstrated that the rate of in-hospitalmortality for patients diagnosedwith takotsubo syndromewas 4.3%. Males and patients with high CCI scores hadhigher rates of mortality than their female and healthiercounterparts. Patients who had underlying criticalillnesses such as sepsis, acute renal failure, acute CVA,respiratory insufficiency, and non-cardiac surgery com-prised most of the patients who had in-hospital mortality—30% of patients who had acute comorbidities com-prised over 80% of patients who had in-hospital mortality.This suggests that the driver of in-hospital mortalityamong patients with takotsubo cardiomyopathy is associ-ated critical illness. Furthermore, male patients had moreunderlying illness than females, which may contribute tothe higher mortality rates that we observed among thisgroup. Further studies are needed to elucidate the causesof increased mortality among male patients and todetermine the demographic, clinical, and imaging pre-dictors of increased mortality and complications forpatients with takotsubo cardiomyopathy.

AcknowledgementsNone.

DisclosuresFinancial support: None.

References1. Dote K, Sato H, Tateishi H, et al. Myocardial stunning due to

simultaneous multivessel coronary spasms: a review of 5 cases.J Cardiol 1991;21:203-14.

2. Pavin D, Le Breton H, Daubert C. Human stress cardiomyopathymimicking acute myocardial syndrome. Heart 1997;78:509-11.

3. Seth PS, Aurigemma GP, Krasnow JM, et al. A syndrome of transientleft ventricular apical wall motion abnormality in the absence ofcoronary disease: a perspective from the United States. Cardiology2003;100:61-6.

4. Gianni M, Dentali F, Grandi AM, et al. Apical ballooning syndromeor takotsubo cardiomyopathy: a systematic review. Eur Heart J 2006;27:1523-9.

5. Charlson ME, Pompei P, Ales KL, et al. A new method of classifyingprognostic comorbidity in longitudinal studies: development andvalidation. J Chronic Dis 1987;40:373-83.

6. Akashi YJ, Musha H, Kida K, et al. Reversible ventricular dysfunctiontakotsubo cardiomyopathy. Eur J Heart Fail 2005;7:1171-6.

7. Eitel I, von Knobelsdorff-Brenkenhoff F, Bernhardt P, et al. Clinicalcharacteristics and cardiovascular magnetic resonance findings instress (takotsubo) cardiomyopathy. JAMA 2011;306:277-86.

8. Kurisu S, Sato H, Kawagoe T, et al. Tako-tsubo-like left ventriculardysfunction with ST-segment elevation: a novel cardiac syndromemimicking acute myocardial infarction. Am Heart J 2002;143:448-55.

9. Madhavan M, Rihal CS, Lerman A, et al. Acute heart failure in apicalballooning syndrome (TakoTsubo/stress cardiomyopathy): clinicalcorrelates and Mayo Clinic risk score. J Am Coll Cardiol 2011;57:1400-1.

Brinjikji et al 221American Heart JournalVolume 164, Number 2

10. Tsuchihashi K, Ueshima K, Uchida T, et al. Transient left ventricularapical ballooning without coronary artery stenosis: a novel heartsyndrome mimicking acute myocardial infarction. Angina Pectoris-Myocardial Infarction Investigations in Japan. J Am Coll Cardiol2001;38:11-8.

11. Yoshimura S, Toyoda K, Ohara T, et al. Takotsubo cardiomyopathyin acute ischemic stroke. Ann Neurol 2008;64:547-54.

12. Roe MT, Messenger JC, WeintraubWS, et al. Treatments, trends, andoutcomes of acute myocardial infarction and percutaneous coronaryintervention. J Am Coll Cardiol 2010;56:254-63.

13. Sharkey SW, Windenburg DC, Lesser JR, et al. Natural history andexpansive clinical profile of stress (tako-tsubo) cardiomyopathy. J AmColl Cardiol 2010;55:333-41.

14. Lee PH, Song JK, Sun BJ, et al. Outcomes of patients with stress-induced cardiomyopathy diagnosed by echocardiography in atertiary referral hospital. J Am Soc Echocardiogr 2010;23:766-71.

15. Geng S, Mullany D, Fraser JF. Takotsubo cardiomyopathy associatedwith sepsis due to Streptococcus pneumoniae pneumonia. Crit CareResusc 2008;10:231-4.

16. Karvouniaris M, Papanikolaou J, Makris D, et al. Sepsis-associatedtakotsubo cardiomyopathy can be reversed with levosimendan. Am JEmerg Med 2012;30:832.e5-7.

17. Sarullo FM, Americo L, Accardo S, et al. Tako-tsubo cardiomyopathyobserved in a patient with sepsis and transient hyperthyroidism.Monaldi Arch Chest Dis 2009;72:33-6.

18. Yonekawa K, Mussio P, Yuen B. Ischemic cerebrovascular stroke ascomplication of sepsis-induced Takotsubo cardiomyopathy. InternEmerg Med 2011;6:477-8.

19. Bradbury B, Cohen F. Early postoperative Takotsubo cardiomyopa-thy: a case report. AANA J 2011;79:181-8.

20. Jabiri MZ, Mazighi M, Meimoun P, et al. Tako-tsubo syndrome: acardioembolic cause of brain infarction. Cerebrovasc Dis 2010;29:309-10.

21. Song BG, Hahn JY, Cho SJ, et al. Clinical characteristics, ballooningpattern, and long-term prognosis of transient left ventricularballooning syndrome. Heart Lung 2010;39:188-95.

22. Elesber AA, Prasad A, Lennon RJ, et al. Four-year recurrence rate andprognosis of the apical ballooning syndrome. J Am Coll Cardiol2007;50:448-52.

23. Ionescu CN, Aguilar-Lopez CA, Sakr AE, et al. Long-termoutcome of Tako-tsubo cardiomyopathy. Heart Lung Circ 2010;19:601-5.

24. Nanda S, Pamula J, Bhatt SP, et al. Takotsubo cardiomyopathy—anew variant and widening disease spectrum. Int J Cardiol 2007;120:e34-6.

25. Haghi D, Athanasiadis A, Papavassiliu T, et al. Right ventricularinvolvement in Takotsubo cardiomyopathy. Eur Heart J 2006;27:2433-9.

26. Fitzgibbons TP, Madias C, Seth A, et al. Prevalence andclinical characteristics of right ventricular dysfunction intransient stress cardiomyopathy. Am J Cardiol 2009;104:133-6.

27. Maeder M, Fehr T, Rickli H, et al. Sepsis-associated myocardialdysfunction: diagnostic and prognostic impact of cardiac troponinsand natriuretic peptides. Chest 2006;129:1349-66.

28. Bybee KA, Prasad A. Stress-related cardiomyopathy syndromes.Circulation 2008;118:397-409.