Little is known about the epidemiology of stress (takotsubo) cardiomyopathy (SC). We used a 3-arm case–control study to assess differences in demographic and co-morbid predictors of SC compared to orthopedic controls and myocardial infarction (MI) controls to characterize (1) population-level predictors of SC generally and (2) differences and similarities in determinants of SC compared to MI. We included data on all discharges of patients diagnosed with SC from the 2008 to 2009 National Inpatient Samples and randomly selected 1-to-1 age-matched controls from patients hospitalized with MI and patients hospitalized with joint injuries after trauma. We used McNemar tests to assess differences in demographic characteristics and co-morbidities between patients with SC and controls. There were 24,701 patients with SC in our study. Of patients with SC, 89.0% were women compared to 38.9% of patients with MI and 55.7% of orthopedic controls. Patients with SC were more likely to be white and to reside in wealthier ZIP codes compared to MI and orthopedic controls. Patients with SC were less likely to have cardiovascular risk factors compared to MI and orthopedic controls but were more likely to have had histories of cerebrovascular accidents, drug abuse, anxiety disorders, mood disorders, malignancy, chronic liver disease, and sepsis. In conclusion, demographic and co-morbid predictors of SC differ substantially from those of MI and may be of interest to providers when diagnosing SC. Several co-morbid risk factors predictive of SC may operate by increased catecholamines.
Stress (takotsubo) cardiomyopathy (SC), also known as “broken heart syndrome,” is a clinical syndrome characterized by acute reversible dysfunction of the middle and/or apical segments of the left ventricle in the context of an acute stressor in patients without culprit coronary disease. Sharing a very similar presentation with myocardial infarction (MI), SC often presents as chest pain and dyspnea sometimes in addition to syncope and palpitations. Most common in women after menopause, investigators have suggested that postmenopausal hormone disequilibrium may play a role in the pathophysiology of the condition. We designed a 3-arm case–control study using data from the 2008 to 2009 National Inpatient Samples (NIS) to meet 2 aims: (1) to characterize demographic and co-morbid determinants of SC overall and (2) to contrast demographic and comorbid predictors of SC from those of MI to educate providers attempting to differentiate these at diagnosis. Case–control studies are particularly useful when attempting to draw inference about the epidemiology of rare diseases such as SC. However, to avoid selection bias in case–control studies, controls must be selected independently of a likelihood of exposure to factors that may be associated with the outcome of interest. This was particularly challenging given the limitations of our dataset, a sample population that was necessarily hospitalized and therefore likely to share common risk factors with SC. To meet our primary aim, therefore, we contrasted demographic and co-morbid profiles of patients with SC with a 1-to-1 random age-matched sample of patients with orthopedic joint injuries secondary to trauma, which we believed, although imperfect, would minimize selection bias. To meet our second aim, we contrasted demographic and co-morbid predictors of SC compared to a 1-to-1 random age-matched sample of patients hospitalized with MI. By directly comparing patients with SC to those with MI, we can differentiate between their demographic and co-morbid profiles, which can be used by clinicians making decisions about diagnosis.
Methods
We purchased the NIS hospital discharge database for 2008 to 2009 from the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality (Rockville, Maryland). The NIS is a hospital discharge database that represents 20% of all inpatient admissions to nonfederal hospitals in the United States. For this analysis, data were weighted by hospital characteristics to ensure national representativeness. We included all patients with a diagnosis of SC (n = 24,701, code 42983 in the International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM]). Sixty-nine percent of these patients had diagnoses confirmed by echocardiogram or cardiac catheterization. We compared demographic and co-morbid predictors for SC to 2 control groups. Our first control group consisted of patients with a primary diagnosis of MI from 2008 through 2009 (n = 25,069) and our second control group consisted of patients with a primary diagnosis of orthopedic joint injury secondary to trauma from 2008 through 2009 (n = 24,601, slight differences in patient counts across case and control populations are due to hospital-level weighting, described earlier). Patients were matched by age across case and control groups to adjust for confounding by age.
Using ICD-9-CM diagnosis codes and Clinical Classifications Software codes, we determined the proportion of patients in each group with the co-morbidities listed below. Clinical Classifications Software (Agency for Healthcare Research and Quality, Rockville, Maryland) coding is a categorization scheme used by the NIS to collapse multiple ICD-9-CM codes into a smaller number of categories to facilitate statistical analysis ( http://www.hcup-us.ahrq.gov/toolssoftware/ccs/ccs.jsp ). Cardiovascular diseases and risk factors considered included obesity, hypertension, hyperlipidemia, diabetes mellitus, coronary artery disease, and smoking. Pulmonary diseases studied included asthma, chronic obstructive pulmonary disease, and pulmonary circulation disorders (pulmonary embolism, history of pulmonary embolism, pulmonary hypertension). Endocrine diseases studied included hyperthyroidism and pheochromocytoma. Neurologic diseases studied included subarachnoid hemorrhage, intracerebral hemorrhage, and cerebrovascular accident (including stroke and transient ischemic attack). Psychiatric diseases studied included drug abuse, alcohol abuse, anxiety disorders, mood disorders, and dementia/delirium. In addition, we considered chronic kidney disease, chronic liver disease, connective tissue disease, sepsis, and malignancy. The Charleston Co-morbidity Index was also calculated for each subject.
For this statistical analysis, we summed data from 2008 through 2009. First, we calculated univariate statistics to describe our sample. Demographic factors we considered included age, gender, race, mean annual household income of ZIP code of residence, and lack of insurance. Second, given the 3-arm case–control design, we calculated odds ratios of exposure in patients with SC compared to each control group separately. McNemar test of association was used to assess differences in exposure to each demographic factor and co-morbidity in matched pairs and odds of exposure and 95% confidence intervals were calculated in cases compared to controls. To compare differences in continuous variables between cases and controls, we performed matched-pair t tests.
Statistical analyses were performed using the SAS-based statistical package JMP 9.0 ( http://www.jmp.com ) and SAS 9.1 ( http://www.sas.com ).
Results
Table 1 presents demographic characteristics of patients diagnosed with SC, age-matched MI controls, and age-matched orthopedic controls and McNemar tests of the association between each demographic factor and outcome separately by control group. Patients with SC were significantly more likely to be women (p <0.001). There were also significant differences by race (p <0.001), income (p <0.001), and insurance status (p <0.001); patients with SC were more likely to be white and have higher incomes than those with MI or orthopedic patients. Patients with SC were more likely to be insured than patients with MI but less likely to be insured than orthopedic patients.
| Variable | All SC | MI | p Value, MI vs SC | Orthopedic | p Value, Orthopedic vs SC |
|---|---|---|---|---|---|
| (n = 24,701) | (n = 25,069) | (n = 24,601) | |||
| Age (years), mean ± SD | 66.9 ± 30.7 | 67.0 ± 30.3 | 0.72 | 67.3 ± 31.4 | 0.21 |
| Age group (years) | 0.99 | 0.99 | |||
| <50 | 2,689 (10.9%) | 2,722 (10.9%) | 2,666 (10.8%) | ||
| 50–64 | 7,290 (29.5%) | 7,363 (29.4%) | 7,286 (29.6%) | ||
| >64 | 14,722 (59.6%) | 14,975 (59.8%) | 14,696 (59.6%) | ||
| Gender | <0.0001 | <0.0001 | |||
| Men | 2,707 (11.0%) | 15,316 (61.1%) | 10,889 (44.3%) | ||
| Women | 21,994 (89.0%) | 9,753 (38.9%) | 13,712 (55.7%) | ||
| Race | <0.0001 | <0.0001 | |||
| White | 16,680 (84.0%) | 16,150 (76.8%) | 16,562 (80.9%) | ||
| Black | 1,178 (5.9%) | 1,905 (9.1%) | 1,444 (7.1%) | ||
| Hispanic | 1,032 (5.2%) | 1,569 (7.5%) | 1,452 (7.1%) | ||
| Asian | 353 (1.8%) | 365 (1.7%) | 301 (1.5%) | ||
| Native-American | 114 (0.6%) | 101 (0.5%) | 145 (0.7%) | ||
| Other | 508 (2.6%) | 950 (4.5%) | 582 (2.8%) | ||
| Income (US$) | <0.0001 | <0.0001 | |||
| <39,000 | 5,196 (21.5%) | 7,339 (30.1%) | 5,758 (24.1%) | ||
| 39,000–62,999 | 12,773 (52.7%) | 12,419 (50.9%) | 12,550 (51.5%) | ||
| >63,000 | 6,255 (25.8%) | 4,627 (19.0%) | 5,543 (23.2%) | ||
| Uninsured | 830 (3.4%) | 1,680 (6.7%) | <0.0001 | 609 (2.5%) | <0.0001 |
Table 2 presents co-morbidities of patients diagnosed with SC, age-matched MI controls, and age-matched orthopedic controls and McNemar tests of the association between each co-morbidity and outcome separately by control group. Compared to orthopedic controls, with respect to cardiovascular risk factors, patients with SC were less likely to be obese, more likely to be hyperlipidemic, more likely to have coronary artery disease, and more likely to smoke. With respect to pulmonology co-morbidities, patients with SC were also more likely to have asthma, more likely to have chronic obstructive pulmonary disease, and substantially more likely to have pulmonary circulation disorders. With respect to endocrine co-morbidities, patients with SC had higher risk for hyperthyroidism. Neurologically, patients with SC had much higher risk for cerebrovascular accidents. Patients with SC also differed on psychiatric co-morbidities. Patients with SC had higher risk for drug abuse and anxiety disorders and lower risk for dementia. With respect to other disorders, patients with SC had higher risk for chronic kidney disease, chronic liver disease, connective tissue disorders, malignancy, and sepsis.
| Frequency (%) | OR (95% CI) | p Value, McNemar Test | |||||
|---|---|---|---|---|---|---|---|
| All SC (n = 24,701) | MI (n = 25,069) | Orthopedic Control (n = 24,601) | SC vs MI | SC vs Orthopedic | SC vs MI | SC vs Orthopedic | |
| Charlson Co-morbidity Index score | 1.4 (2.7%) | 2.4 (3.2%) | 0.6 (2.1%) | <0.0001 | <0.0001 | ||
| Obesity | 1,494 (6.1%) | 2,930 (11.7%) | 2,518 (10.2%) | 0.49 (0.46–0.52) | 0.57 (0.53–0.60) | <0.0001 | <0.0001 |
| Hypertension | 14,434 (58.4%) | 17,458 (69.6%) | 14,117 (57.3%) | 0.61 (0.59–0.64) | 1.05 (1.01–1.09) | <0.0001 | 0.009 |
| Hyperlipidemia | 9,261 (37.5%) | 13,553 (54.1%) | 6,694 (27.2%) | 0.51 (0.49–0.53) | 1.61 (1.55–1.67) | <0.0001 | <0.0001 |
| Diabetes mellitus | 4,661 (18.9%) | 9,282 (37.0%) | 5,245 (21.3%) | 0.40 (0.38–0.41) | 0.86 (0.82–0.90) | <0.0001 | <0.0001 |
| Coronary artery disease | 11,013 (44.6%) | 20,078 (80.1%) | 4,278 (17.4%) | 0.20 (0.19–0.21) | 3.83 (3.68–3.99) | <0.0001 | <0.0001 |
| Smoker | 3,250 (13.2%) | 5,487 (21.9%) | 2,262 (9.2%) | 0.54 (0.52–0.57) | 1.50 (1.42–1.59) | <0.0001 | <0.0001 |
| Asthma | 2,063 (8.4%) | 1,059 (4.2%) | 1,550 (6.3%) | 2.07 (1.91–2.23) | 1.36 (1.27–1.45) | <0.0001 | <0.0001 |
| Chronic obstructive pulmonary disease | 4,615 (18.7%) | 4,035 (16.1%) | 2,169 (8.8%) | 1.20 (1.14–1.25) | 2.38 (2.26–2.51) | <0.0001 | <0.0001 |
| Pulmonary circulation disorder | 1,708 (6.9%) | 941 (3.8%) | 262 (1.1%) | 1.90 (1.75–2.07) | 6.90 (6.05–7.87) | <0.0001 | <0.0001 |
| Hyperthyroidism | 152 (0.6%) | 89 (0.4%) | 54 (0.2%) | 1.73 (1.33–2.25) | 2.81 (2.06–3.84) | <0.0001 | <0.0001 |
| Pheochromocytoma | 20 (0.1%) | 15 (0.1%) | NA | 1.33 (0.69–2.59) | NA | 0.32 | 0.004 |
| Subarachnoid hemorrhage | 223 (0.9%) | NA ⁎ | NA | NA | NA | <0.0001 | <0.0001 |
| Intracerebral hemorrhage | 83 (0.3%) | 35 (0.1%) | NA | 2.43 (1.63–3.61) | NA | <0.0001 | <0.0001 |
| Cerebrovascular accident | 940 (3.8%) | 523 (2.1%) | 90 (0.4%) | 1.86 (1.67–2.07) | 10.81 (8.70–13.43) | <0.0001 | <0.0001 |
| Drug abuse | 862 (3.5%) | 530 (2.1%) | 387 (1.6%) | 1.67 (1.50–1.87) | 2.27 (2.01–2.56) | <0.0001 | <0.0001 |
| Alcohol abuse | 860 (3.5%) | 578 (2.3%) | 764 (3.1%) | 1.53 (1.37–1.70) | 1.13 (1.02–1.24) | <0.0001 | 0.002 |
| Anxiety disorder | 2,204 (8.9%) | 858 (3.4%) | 916 (3.7%) | 2.77 (2.55–3.00) | 2.54 (2.34–2.75) | <0.0001 | <0.0001 |
| Mood disorder | 3,696 (15.0%) | 1,815 (7.2%) | 2,943 (11.9%) | 2.25 (2.13–2.39) | 1.30 (1.23–1.37) | <0.0001 | <0.0001 |
| Delirium/dementia | 1,150 (4.7%) | 1,466 (5.9%) | 1,662 (6.7%) | 0.79 (0.73–0.85) | 0.68 (0.63–0.73) | <0.0001 | <0.0001 |
| Chronic kidney disease | 1,704 (6.9%) | 4,251 (17.0%) | 1,226 (5.0%) | 0.36 (0.34–0.38) | 1.42 (1.31–1.53) | <0.0001 | <0.0001 |
| Chronic liver disease | 1,105 (4.5%) | 790 (3.2%) | 253 (1.0%) | 1.44 (1.31–1.58) | 4.52 (3.94–5.19) | <0.0001 | <0.0001 |
| Connective tissue disease | 282 (1.1%) | 143 (0.6%) | 151 (0.6%) | 2.01 (1.65–2.47) | 1.87 (1.53–2.27) | <0.0001 | <0.0001 |
| Sepsis | 1,761 (7.1%) | 618 (2.5%) | 135 (0.6%) | 3.04 (2.77–3.34) | 13.94 (11.69–16.62) | <0.0001 | <0.0001 |
| Malignancy | 3,547 (14.4%) | 2,506 (10.0%) | 2,176 (8.8%) | 1.51 (1.43–1.59) | 1.73 (1.63–1.83) | <0.0001 | <0.0001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
