We performed a retrospective case-control study using the 2022 National Inpatient Sample (NIS), a component of the Healthcare Cost and Utilization Project (HCUP). The NIS is the largest publicly available, all-payer inpatient database in the United States, capturing approximately 20% of hospital discharges and enabling generation of nationally representative estimates. All adult hospitalizations (≥ 18 years) were eligible for inclusion.

Hospitalizations with TTC were identified using the ICD-10-CM diagnosis code I51.81. Control hospitalizations were selected from the same database as a 1% simple random sample of non-TTC discharges, generated with PROC SURVEYSELECT in SAS using a fixed random seed to ensure reproducibility. A 1% random sample was used to maintain computational feasibility while preserving representativeness of the underlying population through application of survey weights.

The primary exposure was FM, identified using the ICD-10-CM code M79.7, as recorded in any of up to 40 diagnosis fields. ICD-10-CM codes used to define TTC, FM, and all covariates are provided in Table S1.

Multivariable models adjusted for demographics (age, sex, and race/ethnicity) as well as covariates chosen a priori to capture both established TTC risk factors and potential diagnostic confounders. Established risk factors included hypertension, neurologic disease (e.g., stroke, seizures, migraines), and psychiatric disorders (e.g., depression, anxiety, mood disorders). Potential confounders included conditions with overlapping clinical presentations (e.g., chest pain, dyspnea, fatigue) or those contributing to reduced ejection fraction, such as coronary artery disease, atrial fibrillation, chronic kidney disease, obesity, diabetes, cancer, and chronic obstructive pulmonary disease.

To further account for baseline illness severity, we incorporated severity measures from the NIS severity file. Specifically, we used the All Patient Refined Diagnosis Related Group (APR-DRG) severity of illness classification, which categorizes hospitalizations into four levels (minor, moderate, major, and extreme) based on secondary diagnoses and resource utilization. This adjustment helped ensure that the observed association between FM and TTC was not confounded by overall acuity of illness.

Baseline characteristics were compared using chi-square tests for categorical variables and t-tests for continuous variables. Multivariable survey-weighted logistic regression was used to estimate the association between FM and TTC, reporting odds ratios (ORs) with 95% confidence intervals (CIs). Analyses incorporated NIS discharge weights (DISCWT), hospital-level clustering (HOSP_NIS), and stratification variables (NIS_STRATUM) to account for the complex sampling design and produce nationally representative estimates. Because the control group consisted of a 1% random sample of non-TTC hospitalizations, discharge weights for controls were rescaled accordingly to preserve representation of the underlying population. Statistical significance was defined as a two-sided P < 0.05. All analyses were conducted using standard statistical software (SAS).

A total of 9,376 hospitalizations with TTC and 65,690 non-TTC control hospitalizations were included. Patients with TTC were older (mean age 66 vs. 58 years, P < 0.001), more frequently female (86% vs. 61%, P < 0.001), and had a higher prevalence of cardiovascular and psychiatric comorbidities compared with controls (Table 1). FM was present in 282 TTC cases (3.0%) compared with 827 control hospitalizations (1.3%) (P < 0.001).

In multivariable logistic regression, FM was independently associated with increased odds of TTC. After adjusting for demographic characteristics, comorbidities, and APR-DRG severity of illness, FM conferred a 19% increase in odds of TTC (OR 1.19; 95% CI 1.00–1.42, P = 0.044).

This study has limitations that merit consideration.

First, reliance on administrative codes may lead to misclassification of both FM and TTC, as diagnostic accuracy can vary across institutions. The relatively low prevalence of FM observed in both groups likely reflects underdiagnosis or undercoding in administrative datasets, which may result in missed cases. This may lead to nondifferential misclassification of exposure status and underestimation of the true association. Additionally, the use of a 1% sampled control group may introduce selection bias; however, random sampling and application of survey weights were employed to minimize this and preserve representativeness.

Second, the NIS lacks granular clinical data including imaging confirmation, biomarker results, medication exposures, and other physiologic variables. Consequently, TTC diagnoses cannot be independently adjudicated, TTC variants cannot be distinguished, and important clinical contributors—such as precipitating stressors, hemodynamic parameters, or concurrent medication use—cannot be assessed. In addition, FM diagnoses cannot be characterized in terms of severity, chronicity, or symptom burden. These limitations restrict the ability to evaluate underlying mechanisms and introduce potential diagnostic uncertainty. As a result, interpretability is limited, and the observed associations must be understood within the constraints of administrative coding rather than detailed clinical phenotyping.

Third, FM is a heterogeneous condition with fluctuating symptom burden, yet reliance on a binary ICD-10 code cannot capture disease severity or chronicity.

Fourth, because the NIS is a cross-sectional database of hospitalizations, temporal sequencing cannot be established; we cannot determine whether FM preceded TTC or was identified incidentally during hospitalization.

Fifth, residual confounding is possible despite multivariable adjustment, particularly for psychosocial stressors and medications not well captured in administrative data. FM is closely associated with psychiatric comorbidities such as anxiety and depression, which are also established triggers for TTC. This overlap raises the possibility that the observed association may, in part, reflect shared psychiatric and stress-related vulnerability rather than an independent effect of FM alone. Although adjustments were made for these conditions, psychiatric diagnoses may be incompletely captured or misclassified in administrative data, and detailed clinical information including severity, chronicity, treatment status, and active symptom burden is not available. In addition, key psychosocial factors such as acute emotional stressors or prior trauma, which are known contributors to TTC, are not directly measured in the NIS. As such, psychiatric overlap may partially explain the observed association, and the findings should be interpreted with appropriate caution.

Future work should leverage multicenter registries or prospective cohorts with detailed phenotyping to validate FM as a TTC risk factor, clarify autonomic mechanisms, and inform risk stratification in this population.

FM was associated with modestly increased odds of TTC. While the observed effect size was small and should not be interpreted as establishing causality or implying high absolute risk, this hypothesis-generating association highlights the importance of maintaining clinical vigilance in appropriate contexts and warrants further investigation into mechanisms underlying stress-induced cardiomyopathy.