Etiologies, predictors, and economic impact of readmission within 1 month among patients with takotsubo cardiomyopathy

Abstract

Background

Limited data exist on readmission among patients with takotsubo cardiomyopathy (TC), a commonly reversible cause of heart failure.

Hypothesis

We sought to identify etiologies and predictors for readmission among TC patients.

Methods

We queried the National Readmissions Database for 2013–2014 to identify patients with primary admission for TC using ICD‐9‐CM code 429.83. Patients readmitted to hospital within 1 month after discharge were further evaluated to identify etiologies, predictors, and resultant economic burden of readmission. Additionally, we analyzed readmission for TC at 6 months.

Results

We studied 5997 patients admitted with TC, of whom 1.2% experienced in‐hospital mortality. Median age was 67 years, with 91.5% being female. Among survivors, 10.3% were readmitted within 1 month; 25% of the initial 1‐month readmissions occurred within 4 days, 50% within 10 days, and 75% within 20 days from discharge. The most common etiologies for readmission were cardiac (26%), respiratory (16%), and gastrointestinal (11%) causes. Heart failure was the most common cardiac etiology. Significant predictors of increased 1‐month readmission included systemic thromboembolic events, length of stay ≥3 days, and underlying psychoses. Obesity and private insurance predicted lower 1‐month readmission. The annual national cost impact for index admission and 1‐month readmissions was ≈$112 million. Recurrent TC was seen among 1.9% of patients readmitted within 6 months.

Conclusions

Though the overall rate of 1‐month readmission following TC is low, associated economic burden from readmission is still significant. Patients are readmitted mostly for noncardiac causes. Readmission for another episode of TC within 6 months was uncommon.

1. INTRODUCTION

Takotsubo cardiomyopathy (TC) is known to be a potentially reversible cause of acute heart failure (HF).1 Over the last few years, HF readmissions have been heavily scrutinized, largely due to their significant economic impact on the healthcare system. Furthermore, early hospital readmission after TC has been associated with decreased long‐term survival.2 Various efforts to decrease HF hospital readmissions have been researched and introduced across the country.3 TC is one potential target to help optimize patient outcomes, including hospital readmissions. Considering the reversible natural history of TC, targeting patients with certain underlying risk factors may have significant potential in improving outcomes within this population. However, at the present time, data specifically addressing the plethora of etiologies that are responsible for readmission in patients with TC is limited. The primary objective of this study is to evaluate specific etiologies, predictors, and comorbidities associated with 1‐month or short‐term hospital readmission in patients admitted with TC from one of the largest nationwide readmission databases.

2. METHODS

The study cohort was derived from the Healthcare Cost and Utilization Project's (HCUP) National Readmission Database (NRD) of 2013 and 2014, sponsored by the Agency for Healthcare Research and Quality (AHRQ). The NRD is one of the largest publicly available all‐payer inpatient databases in the United States, including data on approximately 28 million discharges in year 2013 and estimating >50 million discharges from 21 states with reliable, verified linkage numbers. NRD represents 49.1% of total US hospitalizations. Patients were tracked using variable “NRD_visitlink,” and time between 2 admissions was calculated by subtracting variable “NRD_DaysToEvent.” Time to readmission was calculated by subtracting length of stay (LOS) of index admissions to time between 2 admissions. National estimates were produced using sampling weights provided by the sponsor. Details regarding the NRD data are available online (NRD Description of Data Elements, https://www.hcup-us.ahrq.gov/db/vars/nrd_daystoevent/nrdnote.jsp).

Patients admitted with a primary diagnosis of TC identified using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD‐9‐CM) code 429.83 represented the index admissions. We excluded patients age < 18 years and those with missing data for age, sex, or mortality. We also excluded index admissions in the month of December, as we did not have 1‐month follow‐up data for the same. One‐month readmission was defined as any readmission that occurred between 0 and 31 days from hospital discharge.

The primary outcome was 1‐month readmission, which was further divided into readmissions. Readmissions causes were identified by using ICD‐9‐CM codes in the primary diagnosis field. We combined the ICD‐9‐CM codes with similar diagnoses to make clinically important groups. Within the group with primary admissions for TC and readmissions, we use ICD‐9‐CM codes to define additional variables and comorbidities (see Supporting Information, Appendix Table 1, in the online version of this article). Recurrence of TC at 6 months was calculated specifically among index admissions that occurred between January and June for both years of study.

NRD variables were used to identify patients' demographic characteristics including age and sex; hospital characteristics such as bed size and teaching status; and other patient‐specific characteristics of primary payer, admission type, admission day, and discharge disposition. “CM_” variables identified different comorbidities by using ICD‐9‐CM diagnoses and the diagnosis‐related group in effect on the discharge date. These comorbidities are not directly related to the principal diagnosis or the main reason for admission and are likely to have originated before the hospital stay. The severity of comorbid conditions was defined using Quan modification of the Charlson Comorbidity Index (CCI), which contains 17 comorbid conditions with differential weights. The score ranges from 0 to 33, with higher scores corresponding to greater burden of comorbid diseases.4

2.1. Statistical analysis

All analyses were performed using SPSS version 23.0 (IBM Corp., Armonk, NY) on unweighted NRD data using similar statistical methods as previously published.5, 6, 7 Differences between categorical variables were tested using the χ² test and continuous variables by using the Student t test. The annual cost impact of index and 1‐month readmissions was analyzed using weighted samples to estimate national averages. A multivariable hierarchical mixed‐effects logistic regression model was created to assess which variables predicted 1‐month readmission. The multivariate models for readmission included hospital‐level variables such as bed size and teaching status; patient‐level variables like age group, sex, CCI, admission type, admission day, primary payer; LOS of index admission; calendar year; comorbidities; and disposition post–index admission (see Supporting Information, Appendix Table 3, in the online version of this article).

3. RESULTS

3.1. Baseline characteristics

The data we analyzed included a total of 5997 (weighted n = 13 620) index admissions for TC in the United States during the prespecified study period (Table). The study population was predominantly female, at 91.5%. In‐hospital mortality among index admissions was 1.2%. Among the survivors of index admission, 612 (weighted n = 1464) patients were readmitted within 1 month, representing a 10.3% readmission rate.

The patients who were readmitted had a higher burden of significant comorbidities at baseline, reflected by their higher CCI of ≥2 compared with patients who were not readmitted (62.3% vs 38.9%, respectively). Patients who were readmitted had significantly higher rates of baseline hypertension (HTN; 75.3% vs 64.9%), diabetes mellitus (26.6% vs 19.9%), chronic pulmonary disease (40% vs 25.9%), stroke or transient ischemic attack (10.6 vs 6.3%), peripheral vascular disease (12.9% vs 6.9%), deficiency anemias (19% vs 10.7%), congestive HF (CHF; 10% vs 4.4%), hypothyroidism (23% vs 17.5%), and renal failure (12.3% vs 6.5%) compared to patients who were not readmitted. A majority of the readmitted patients were from large hospitals (a large number of beds; 66.3%), with the primary insurance payer being Medicare at 71.4% (Table 1). The rate of systemic thromboembolic events was 1.6% within the index admission group (excluding myocardial infarction and stroke) and 4.5% among all readmissions that occurred within 1 month post‐discharge.

Table 1.

Baseline characteristics of patients during index admission

Characteristics	Readmission		P Value
	Yes, n = 612	No, n = 5385
Mean age, y	69.4 ± 13.0	66.4 ±12.8	<0.001
Female sex	91.2	91.5	0.78
Administrative data
Weekend admission	24.7	24.5	0.93
Elective admission	3.1	3.7	0.47
Payer			<0.001
Medicare	71.4	58.9
Medicaid	7.7	6.7
Private	16.5	28.0
Self‐pay	2.8	3.2
No charge	0.3	0.6
Other	1.3	2.6
Cost of hospitalization, USD, mean	17 430	13 499	<0.001
LOS, d	5.5 ± 5.1	3.6 ± 3.7	<0.001
Hospital size (based on no. of beds)			0.42
Small	7.0	8.6
Medium	26.6	26.0
Large	66.3	65.4
Hospital location/teaching status			0.06
Urban nonteaching	30.7	34.1
Urban teaching	66.7	62.2
Rural	2.6	3.8
Year			0.78
2013	47.9	48.5
2014	52.1	51.5
Patient comorbidities
CCI			<0.001
≤1	37.7	61.1
2	28.6	20.3
≥3	33.7	18.6
HTN, with and without complications	75.3	64.9	<0.001
DM, with and without complications	26.6	19.5	<0.001
Dyslipidemia	50.5	50.9	0.83
Chronic pulmonary disease	40.0	25.9	<0.001
Pulmonary circulation disorders	1.8	1.1	0.11
Current or past smoker	35.8	35.4	0.83
History of stroke/TIA	10.6	6.3	<0.001
History of MI	8.5	6.9	0.13
Drug abuse	5.4	3.6	0.02
Alcohol abuse	5.1	3.5	0.04
PVD/disorders	12.9	6.9	<0.001
Coagulopathy	4.4	2.8	0.02
Deficiency anemia	19.0	10.7	<0.001
Chronic blood‐loss anemia	0.7	0.4	0.38
Collagen vascular disease or RA	5.7	4.4	0.13
CHF	10.0	4.4	<0.001
Hypothyroidism	23.0	17.5	0.001
Liver disease	3.1	1.7	0.01
Fluid and electrolyte disorders	39.7	23.8	<0.001
Obesity	9.2	10.9	0.17
Renal failure	12.3	6.5	<0.001
Depression	18.3	15.5	0.07
Psychoses	9.3	4.2	<0.001
Valvular disease	2.6	1.7	0.09
Sepsis	3.6	1.9	0.005
Cardiogenic shock	5.7	4.2	0.08
Vasopressor use	1.5	0.9	0.14
Acute cardiorespiratory failure	37.6	22.4	<0.001
Ventilator use	6.0	4.8	<0.181
Ischemic stroke	1.6	0.7	0.02
TIA	0.7	0.3	0.21
Systemic TE event	3.6	1.4	<0.001
Long‐term anticoagulant use	7.4	4.4	0.001
Long‐term antiplatelet use	12.9	13.4	0.73
Disposition
Transfer to SNF/ICF	19.6	7.1	<0.001
In‐hospital mortality	0.0	1.3	N/A

Abbreviations: CCI, Charlson Comorbidity Index; CHF, congestive heart failure; DM, diabetes mellitus; HTN, hypertension; ICF, intermediate‐care facility; LOS, length of stay; MI, myocardial infarction; N/A, not applicable; PVD, peripheral vascular disease; RA, rheumatoid arthritis; SD, standard deviation; SNF, skilled‐nursing facility; TE, thromboembolism; TIA, transient ischemic attack; USD, US dollars.

Data are presented as % or mean ± SD.

3.2. Etiologies and predictors for 1‐month readmission

There were 543 patients with a single readmission, 63 patients had 2 readmissions, 5 patients had 3 readmissions, and 1 patient had 4 readmissions. A total of 24/612 (3.9%) in‐hospital deaths occurred during readmission within 1 month post‐discharge from index hospitalization for TC. The trends for time to initial readmission within the first 1 month revealed that 25% of readmissions occurred within 4 days, 50% occurred within 10 days, and > 75% of the readmissions occurred within 20 days from discharge (Figure 1).

Figure 1.

Time to readmission post‐discharge from index admission

Among the etiologies for readmission, the most common causes for readmission were cardiac (25.6%) and respiratory (16.4%), followed by gastrointestinal (10.7%). Among the specific cardiac causes, HF was the most common (41%), followed by TC (20.7%), arrhythmias (16.8%; especially atrial fibrillation/flutter), and acute coronary syndromes (15.2%). The respiratory causes of readmission were primarily driven by chronic pulmonary disease, respiratory failure, and pneumonia (Figure 2).

Figure 2.

Charts showing (A) all etiologies for readmission following index admission and (B) specific cardiac etiologies for readmission following index admission

Significant predictors of increased 1‐month readmission included LOS ≥8 days (adjusted odds ratio [OR]: 2.12, 95% confidence interval [CI]: 1.48–3.05, P < 0.01), systemic thromboembolic event (OR: 1.82, 95% CI: 1.05–3.15, P < 0.03), psychoses (OR: 1.9, 95% CI: 1.36–2.66, P < 0.01), and additional comorbidities such as HTN, chronic pulmonary disease, and CHF, among others (Figure 3 and Supporting Information, Appendix Table 2, in the online version of this article). Transfer to a skilled nursing facility or intermediate‐care facility yielded greater odds of 1‐month readmission (OR: 1.45, 95% CI: 1.11–1.91, P < 0.01). Private insurance payer and presence of obesity predicted lesser odds for 1‐month readmission.

Figure 3.

Significant predictors for readmission at 1 month among those admitted primarily for takotsubo cardiomyopathy. Abbreviations: AOR, adjusted odds ratio; LCL, 95% confidence interval lower limit; LOS, length of stay; UCL, 95% confidence interval upper limit

3.3. Cost impact of 1‐month readmission

The mean cost per weighted index admission was $13 659, amounting to a $101 million annual cost within the United States. The cost for readmissions amounted to roughly $11 million annually. Readmissions accounted for an estimated 10% of the total costs involved (readmission cost / readmission + index admission costs).

3.4. Readmission at 6 months

Among 3046 index admissions that occurred between January and June, 667 patients (21.9%) were readmitted at least once within 6 months. Cardiac causes comprised 21.3% of readmissions within 6 months, with acute HF being the most common cardiac cause (7.3% of all readmissions). There were 38 (5.7%) in‐hospital deaths among these 667 patients within 6 months. Readmission for a primary diagnosis of TC occurred among 13/667 (1.9%) of the readmitted patients at 6 months. A total of 18 readmissions for TC occurred among these 13 patients. The initial readmission for TC occurred within the first 18 days of discharge from index admission in 12 of the 13 patients.

4. DISCUSSION

Previous studies on rehospitalization for patients with TC have been significantly limited by their small sample sizes, their single‐center nature, and the lack of focus on factors that predict readmissions.8, 9 One study found that recent hospital admission prior to TC diagnosis and increased length of index hospitalization were associated with higher early rehospitalization rate after TC.2 No study so far has looked into the specific comorbidities that contribute to readmissions among patients admitted with TC. We report recent data from one of the largest national databases on the etiologies and predictors of readmission for patients with TC.

Our study population was predominantly female (91.5%), with a prevalence consistent with the international Takotsubo Registry, which was close to 90% female.10 One‐month readmission rates were comparable with a previous registry analysis of 11.6%.8 However, the absolute rate of 1‐month readmission was significantly lower compared with what has been noted among studies evaluating other cardiovascular disorders.6, 11, 12, 13 Factors that may have contributed to this phenomenon are the low overall comorbidity burden, as evidenced by the lower mean CCI score, and an overall lesser severity of illness among TC patients in our study compared with these studies. To our knowledge, no published studies have looked specifically into the comorbidities associated with readmission among TC patients. We found that patients with TC with 1‐month readmissions had significantly higher comorbidities, namely HTN, diabetes, chronic pulmonary disease, stroke/TIA, peripheral vascular disease, iron‐deficiency anemia, CHF, hypothyroidism, and renal failure at baseline.

TC is unique in that the majority of the patients recover their left ventricular ejection fraction and previous cardiovascular status within 3 to 9 days.14 This is significant because the most common causes for readmissions among patients with TC were noncardiac in nature in >75% of the cases. According to our analysis, HF accounted for approximately 10% of readmissions overall. These findings are in contrast to readmission among HF patients, where cardiac causes accounted for half and HF for more than one‐third of all readmissions within 30 days post‐discharge.5 Previous studies evaluating readmission among patients admitted for acute coronary syndrome and patients undergoing percutaneous coronary intervention have also demonstrated higher rates of readmission for cardiac‐related disorders.15

The median age of 67 years in our study is consistent with prior literature. Older age (≥75 years) is known to be associated with HF, stroke, in‐hospital adverse events, and higher in‐hospital mortality.16 Age was not found to be a predictor of 1‐month hospital readmissions on multivariate adjusted analysis; however, our baseline trends were consistent with older patients being incrementally more likely to be readmitted. Patients in the oldest age category (≥80 years) were more than 2‐fold as likely to get readmitted within 1 month compared with patients age < 50 years. We found that longer LOS was associated with higher rates of 1‐month hospital readmissions. Longer LOS has been shown to predict readmission in previous large‐scale observational studies examining inpatients within the United States.17, 18 Patients who require longer hospital stays are commonly sicker than those who require shorter hospital stays, exposing them to a higher risk for early hospital readmission.

Factors predicting readmission such as LOS, chronic pulmonary disease, peripheral vascular disease, and transfer to a facility have been shown to predict higher readmission among HF patients, which is similar to our findings among TC hospitalizations despite the differences between the 2 groups at baseline.5 Such a finding could mean that strategies employed to reduce HF readmission may have a role in preventing post‐TC readmission; however, studies are needed to test the hypothesis. Discharge to skilled nursing or an intermediate‐care facility predicted higher 1‐month hospital readmission. Past studies have shown that for patients with HF, discharge to a skilled nursing facility was associated with a higher risk of death and readmissions.19 This finding could be explained by the fact that patients discharged to skilled nursing or an intermediate‐care facility are usually sicker than patients discharged to home. Obesity was shown to be predictive of lower 1‐month hospital readmission, according to our analysis. HF in obesity is associated with increased left ventricular remodeling20 and pronounced neurohormonal activity,21, 22 both of which may be related to the pathogenesis of TC. However, lower adjusted readmission rates early on after discharge may be an extension of the “obesity paradox” associated with better short‐term outcomes among obese patients when compared with patients with normal body mass index.23 One of the possible explanations of this phenomenon is that body mass index is not the best indicator of body‐fat composition.24 Presence of cardiogenic shock or vasopressor/inotrope use during index hospitalization did not predict readmission among survivors in our study. Cardiogenic shock was noted in 53.5% of the TC patients who died at the end of the index admission but only 3.5% among survivors. The lack of predictability of 1‐month readmission by these factors may be explained by the low rate of both among survivors of index hospitalization and the possibility that survivors may have had lesser severity of disease. Close follow‐up, early outpatient visits, monitoring closely for recovery of ventricular function, patient compliance with appropriate medical management, and improved awareness at the patient and care‐team level could prevent further morbidity within this population, but this requires further study.

Recurrence of TC has been studied in several prior analyses. A meta‐analysis reviewing 298 studies revealed recurrence of the disease in roughly 1.5% of the patients at 1 year and 5% at 6 years.25 We found that ≈0.4% of the patients needed readmission for TC within 6 months following index hospitalization. More than 90% of TC recurrences occurred within the first 3 weeks. Restriction of our analysis to the primary diagnosis for readmission, combined with our inability to measure emergency or outpatient visits, may have potentially underestimated the actual recurrence rate.

The cost impact of TC was noted to be significant, with an annual nationwide cost burden nearing >$112 million for the initial admission and readmission within 1 month. One‐month readmissions accounted for 10% of all the total costs, posing a significant economic burden. The marked cost impact of readmissions as noted in previous studies and the associated penalties executed by the Centers for Medicare and Medicaid Services (CMS) remain at the center of driving forces directed toward reducing readmissions.5

4.1. Study limitations

The strengths of our study include the large sample of patients hospitalized with TC. Further, because it utilized administrative data obtained with appropriately validated methodology, it can be potentially free of selection bias. However, our study is not without limitations. We only looked for potential predictors of hospital readmission on a select number of comorbidities. There are other factors that may have affected patients' prognosis after hospital discharge: discharge medications, including β‐blockers, angiotensin‐converting enzyme inhibitors, or anticoagulants; dietary compliance and volume status; information regarding left and right ventricular function; and recovery to normal cardiac function. Despite the inherent limitations in our study, we believe that the large sample size and real‐world data presented will contribute to the literature on TC.

5. CONCLUSION

Our study is novel in that it is the first large‐scale study that identifies risk factors predictive of readmission following admission for TC. Further, we identified that readmissions are common and primarily for noncardiac causes. Readmission for recurrent TC was present among 0.4% of the index TC patients at 6 months and occurs predominantly within 3 weeks from discharge. Awareness of the predictors for readmission and its causes can help devise strategies to reduce readmissions and costs.

Author contributions

Mahek Shah, MD, and Pradhum Ram, MD, contributed equally. All authors had access to the data and a role in writing the manuscript. The authors are solely responsible for the study design, conduct, and analyses, and the drafting and editing of the manuscript and its final contents.

Conflicts of interest

The authors declare no potential conflicts of interest.

Supporting information

Appendix Table 1. ICD‐9CM codes used for diagnoses and procedures.

Appendix Table 2. Etiologies for one‐month readmission following primary admit for Takotsubo cardiomyopathy.

Appendix Table 3. Multivariable predictors for readmission within 1‐month

Shah M, Ram P, Lo KBU, et al. Etiologies, predictors, and economic impact of readmission within 1 month among patients with takotsubo cardiomyopathy. Clin Cardiol. 2018;41:916–923. 10.1002/clc.22974