Association between insurance status and in‐hospital outcomes in patients with out‐of‐hospital ventricular fibrillation arrest

Abstract Background Lack of health insurance is associated with adverse clinical outcomes; however, the association between health insurance status and in‐hospital outcomes after out‐of‐hospital ventricular fibrillation (OHVFA) arrest is unclear. Hypothesis Lack of health insurance is associated with worse in‐hospital outcomes after out‐of‐hospital ventricular fibrillation arrest. Methods From January 2003 to December 2014, hospitalizations with a primary diagnosis of OHVFA in patients ≥18 years of age were extracted from the Nationwide Inpatient Sample. Patients were categorized into insured and uninsured groups based on their documented health insurance status. Study outcome measures were in‐hospital mortality, utilization of implantable cardioverter defibrillator (ICD), and cost of hospitalization. Inverse probability weighting adjusted binary logistic regression was performed to identify independent predictors of in‐hospital mortality and ICD utilization and linear regression was performed to identify independent predictors of cost of hospitalization. Results Of 188 946 patients included in the final analyses, 178 005 (94.2%) patients were insured and 10 941 (5.8%) patients were uninsured. Unadjusted in‐hospital mortality was higher (61.7% vs. 54.7%, p < .001) and ICD utilization was lower (15.3% vs. 18.3%, p < .001) in the uninsured patients. Lack of health insurance was independently associated with higher in‐hospital mortality (O.R = 1.53, 95% C.I. [1.46–1.61]; p < .001) and lower utilization of ICD (O.R = 0.84, 95% C.I [0.79–0.90], p < .001). Cost of hospitalization was significantly higher in uninsured patients (median [interquartile range], p‐value) ($) (39 650 [18 034‐93 399] vs. 35 965 [14 568.50‐96 163], p < .001). Conclusion Lack of health insurance is associated with higher in‐hospital mortality, lower utilization of ICD and higher cost of hospitalization after OHVFA.

cost of hospitalization. Inverse probability weighting adjusted binary logistic regression was performed to identify independent predictors of in-hospital mortality and ICD utilization and linear regression was performed to identify independent predictors of cost of hospitalization.
Results: Of 188 946 patients included in the final analyses, 178 005 (94.2%) patients were insured and 10 941 (5.8%) patients were uninsured. Unadjusted in-hospital mortality was higher (61.7% vs. 54.7%, p < .001) and ICD utilization was lower (15.3% vs. 18.3%, p < .001) in the uninsured patients. Lack of health insurance was independently associated with higher in-hospital mortality (O.R = 1.53, 95% C.I. Conclusion: Lack of health insurance is associated with higher in-hospital mortality, lower utilization of ICD and higher cost of hospitalization after OHVFA.

| Study population
From January 2003 to December 2014, hospitalizations with a primary diagnosis of VF arrest in patients 18 years of age and older were extracted by searching for the ICD-9-CM codes for VF (427. 4, 427.41, 427.42, and 427.5). Patients with missing data on primary payer status and in-patient mortality were excluded from the final analyses. Figure 1 demonstrates data extraction and patient selection methods.
The primary payer status in the NIS database has been categorized as Medicare, Medicaid, private insurance, self-pay, no charge, and other insurance. If the primary payer status indicated self-pay or no charge, those patients were considered to be uninsured. Patients' baseline comorbidities and procedural characteristics were extracted using ICD-9-CM and CCS codes (Supplemental Table 1 Age, gender, race, prior stroke, diabetes, hypertension, atrial arrhythmias (atrial fibrillation or atrial flutter), chronic kidney disease (CKD), valvular heart disease, long-term use of anti-coagulants, smoking, alcoholism, drug abuse, congestive heart failure, peripheral vascular disease, previous myocardial infarction (MI), previous coronary revascularization with either percutaneous coronary intervention (PCI) or coronary artery bypass graft (CABG) surgery, overweight status, obesity, morbid obesity, and income status were included in the binary logistic regression analysis used to derive probability value for each hospital discharge used as the propensity score. We performed inverse probability weighted analysis using binary logistic regression adjusting for propensity score and distal variables to identify the independent predictors of in-hospital mortality. We also sought to analyze the independent predictors of ICD prescription and assess the effect of insurance status on ICD utilization using a similar method.   mortality and discharge AMA status were associated with lower cost of hospitalization. Table 5  The cost of hospital care for uninsured patients was significantly higher compared to that insured counterparts. This difference was not seen after adjusting for some of the other contributors to the cost of care such as intensive care therapies including mechanical ventilation, and need for mechanical circulatory support which were higher in uninsured patients suggesting these to be the mediators of higher unadjusted cost of hospitalization seen in the uninsured patients.
Higher utilization of these treatments could well be a reflection of the unknown comorbid complexity of uninsured patients.
The higher incidence of behaviors such as history of drug abuse, as well as a higher prevalence of leaving against medical advice, imply a lower level of health awareness and care engagement amongst the uninsured cohort. Healthcare policy interventions to address socioeconomic determinants of health, improving access to health insurance. Enhancing patient education and engagement in both primary care and hospital settings may have potential for improving outcomes of catastrophic presentations such as OHVFA.

| LIMITATIONS
Our observational study from the nation's largest all-payer administrative hospital database has limitations and biases. Being an administrative database, the NIS is subject to coding errors and residual confounding exists. Lack of procedural, as well as laboratory and pharmacotherapy data, further limit the analyses. Since NIS records individual hospitalizations, long-term outcomes impact of insurance status on out-of-hospital VF arrest cannot be analyzed. Our analyses are limited to patients with OHVFA and hence should not be extrapolated to cardiac arrests associated with asystole or PEA.

| CONCLUSIONS
Uninsured victims of out-of-hospital VF arrest have a higher adjusted in-hospital mortality compared to those with health insurance.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from Health Care Utilization Project. Restrictions apply to the availability of these data, which were used under license for this study. Data are available with the permission of Health Care Utilization Project.