Proper Utilization of the National Inpatient Sample Database to Analyze Safety of Cardiovascular Procedures

ImportanceAnalysis of the National Inpatient Sample (NIS) database for the procedural safety of interventions that are primarily outpatient in nature can be inherently biased due to the selective hospitalization of patients who suffer complications during their outpatient procedures. ObjectiveUsing atrial fibrillation (AF) ablation as an example of a procedure typically performed in the outpatient setting, we will test the hypothesis that the population of patients who had their procedure prior to, or on the day of hospital admission (some being outpatient cases, eventually admitted to hospital) will show a significantly higher rate of procedural complications, compared to patients who had a true inpatient ablation, from day 2 of hospital stay and later. DesignThis study is a retrospective analysis of hospitalizations in the NIS dataset, including patients aged 18 and older, who underwent an AF ablation between 2016-2022. Patients were divided into two groups based on the day of hospitalization on which the ablation was performed - day 1 of admission or prior (group 1), vs. day 2 of admission or later (group 2). Patient baseline characteristics and procedural complication rates were compared. SettingHospitals that performed AF ablations in the United States between January 1, 2016, and December 31, 2022. ParticipantsA weighted selection of patients aged 18 years and older who underwent an in-patient AF ablation, as identified by the NIS database (N = 47,925). ExposuresCatheter ablation for AF. Main Outcome(s) and Measure(s)In-hospital procedural complications, length of stay and mortality ResultsThe analysis included 47,925 weighted hospitalizations in patients that underwent AF ablation. Most patients (n= 33,925, 70.8%) underwent catheter ablation (CA) on the day of admission or the days before (group 1). Group 1 was younger, with fewer octogenarians, 6.9% vs. 10.4% (p<0.001), higher proportion of males (60.2% v 58.7%, p=0.003), and had significantly lower rates of comorbid diabetes (24.6% vs. 29.8%), renal disease (12.6% vs. 20.0%), HTN (75.5% vs. 80.1%), heart failure (28.3% vs. 45.8%) and ischemic heart disease (16.0% vs. 20.5%), (p<0.001 for all) compared to Group 2. The rate of complication was higher in group 1 (9.5% v 6.6%, p<0.001). Multivariate analysis showed study group 1 to be an independent predictor of complications (OR 1.54, 95% CI 1.30-1.83, p<0.001). Conclusions and RelevancePatients undergoing their procedure on the day of admission or prior were younger and had significantly fewer comorbidities, consistent with a population that includes outpatient profile characteristics. Despite the younger and healthier population, these patients suffered from a significantly higher rate of procedural complications. The most plausible explanation to these results is a selection bias, where patients undergoing an outpatient procedure who are at higher risk, or develop complications are admitted as inpatients, thus falsely inflating the complication rate among the patients in the NIS database. The NIS database in its current form, has important limitations when used to assess procedures that are primarily outpatient in nature, requiring careful interpretation of the complication rates arising from such analyses.