Atrial fibrillation catheter ablation complications in obese and diabetic patients: Insights from the US Nationwide Inpatient Sample 2005–2013

Abstract Background Obesity and diabetes are risk factors for atrial fibrillation (AF) incidence and recurrence after catheter ablation. However, their impact on post‐ablation complications in real‐world practice is unknown. Objectives We examine annual trends in AF ablations and procedural outcomes in obese and diabetic patients in the US and whether obesity and diabetes are independently associated with adverse outcomes. Methods Using the Nationwide Inpatient Sample (2005–2013), we identified obese and diabetic patients admitted for AF ablation. Common complications were identified using ICD‐9‐CM codes. The primary outcome included the composite of any in‐hospital complication or death. Annual trends of the primary outcome, length‐of‐stay (LOS) and total‐inflation adjusted hospital charges were examined. Multivariate analyses studied the association of obesity and diabetes with outcomes. Results An estimated 106 462 AF ablations were performed in the US from 2005 to 2013. Annual trends revealed a gradual increase in ablations performed in obese and diabetic patients and in complication rates. The overall rate of the primary outcome in obese was 11.7% versus 8.2% in non‐obese and 10.7% in diabetic versus 8.2% in non‐diabetic patients (p < .001). Conclusions Obesity was independently associated with increased complications (adjusted OR, 95% CI:1.39, 1.20–1.62), longer LOS (1.36, 1.23–1.49), and higher charges (1.16, 1.12–1.19). Diabetes was only associated with longer LOS (1.27, 1.16–1.38). Obesity, but not diabetes, in patients undergoing AF ablation is an independent risk factor for immediate post‐ablation complications and higher costs. Future studies should investigate whether weight loss prior to ablation reduces complications and costs.


| INTRODUCTION
Atrial fibrillation (AF) is the most commonly diagnosed cardiac arrhythmia, predicted to affect 6-12 million people in the US by 2050 and 17.9 million in Europe by 2060. 1 It is associated with increased risk of death, heart failure, hospitalization, thromboembolic events and an overall impaired quality-of-life. 1,2 Catheter ablation is a potent rhythm control strategy for patients with symptomatic AF 3 and has been shown to reduce mortality in patients with HF and reduced EF 4 and improve quality-of-life. 5 Data from clinical trials or academic centers report overall complication rates between 2% and 10% and its safety profile has been improving over the years. [6][7][8][9] Obesity is a risk factor for the development of AF due to proposed mechanisms such as left atrial enlargement, inflammation and pericardial fat deposition. 10,11 Diabetes mellitus also is a risk factor, presumably due to increased left ventricular mass and arterial stiffness. 12 Studies have shown that obesity and diabetes, particularly when poorly managed, are associated with high rates of AF recurrence after ablation. [13][14][15] Others have shown that risk factor modification reduces the risk of incident and recurrent AF. 16,17 In terms of procedural complications, meta-analyses of AF ablation data in obese and diabetic patients primarily from specialized centers have shown comparable rates to non-obese and non-diabetic patients. 18,19 However, the real-world rate of adverse in-hospital ablation outcomes in obese and diabetic patients remains largely unknown.
In this study, we aim to use real-world data from the US Nationwide Inpatient Sample (NIS) to determine annual AF ablation rates in obese and diabetic patients and the independent association of these risk factors with post-ablation adverse outcomes.

| Data source
We used the NIS database to obtain patient data from 2005 to 2013. Since the NIS is an administrative and de-identified database, it does not require institutional review board approval or exempt determination and informed consent is waived. The NIS is a large publicly available all-payer inpatient healthcare database designed by the Healthcare Cost and Utilization project to produce US regional and national estimates of inpatient utilization, access, charges, quality, and outcomes accounting for 20% of all US community hospitals. Each entry contains information on demographic details, including age, sex, race, insurance status, primary and secondary procedures, hospitalization outcome, total charges, and length of stay. The NIS contains clinical and resource use information, with safeguards to protect the privacy of patients, physicians, and hospitals. 20,21 Each admission has its own unique identifiers, so repeat ablations or admissions for the same patient could not be distinguished. With the introduction of the two-midnight rule by the Centers for Medicare and Medicaid Services (CMS) in 2013, and its full implementation in 2015, most AF ablations have transitioned to observation or outpatient status. 22 Therefore, we censored our analysis at 2013.

| Study population
We used ICD-9-CM (International Classification of Diseases, Ninth  23,24 Obese and diabetic (combined uncomplicated diabetes and diabetes with chronic complications) patients were identified using the Agency for Healthcare Research and Quality comorbidity measure for ICD-9 codes. ICD-9-CM codes for body mass index (BMI) were not available for the vast majority of admissions and thus were not used to identify or quantitate obesity given concern for sampling error.

| Variables
We used the NIS data elements to identify patient age, sex, race, comorbidities, disposition, hospital volume status, hospital bed-size, and total admission charges. We defined hospital procedural volume as follows: low volume as <50 AF ablations annually, medium volume as 50-100 AF ablations annually, high volume as >100 AF ablations annually. All hospital charges were inflation adjusted using the price index for the gross domestic product and are reported in 2020 US dollars.

| Outcomes
Our primary outcome was the composite of any complication or inhospital death. We included vascular complications (post-op hemorrhage with and without transfusion, blood vessel injury, accidental   punctures, AV fistula, Table S1). All of these complications were selected based on a review of pertinent clinical literature and identified from corresponding ICD-9-CM diagnosis and procedure codes as used by other investigators to maintain consistency, 23,24 although we added more complications like diaphragmatic injury and pericarditis. Secondary outcomes included in-hospital mortality alone, length-of-stay (LOS), and total inflation-adjusted hospital charges. We compared these outcomes in obese and non-obese patients, diabetic and non-diabetic patients and across four subgroups: non-obese non-diabetic, obese non-diabetic, non-obese diabetic, and obese diabetic.

| Statistical analysis
We used the NIS weights to generate national estimates of the number of admissions each year. We used the χ 2 test to compare categorical variables for demographics across the four subgroups of non-obese non-diabetic, obese non-diabetic, diabetic non-obese and obese diabetic, and for comparing the primary outcome between obese versus non-obese and diabetic versus non-diabetic patients.
We used one-way ANOVA for age. Generalized linear models were used to analyze annual trends. We also generated nested multivariable models to analyze independent predictors for our outcomes adjusting for different combinations of age, sex, race, hypertension, renal failure, chronic pulmonary disease, peripheral vascular disease, obesity, diabetes, hospital procedure volume, and bed-size. We used logistic regression used for the primary outcome of any complication including death and secondary outcomes of in-hospital mortality alone and LOS (0-1 vs. ≥2 days). Linear regression was used for inflation-adjusted charges, which were initially log transformed.  Significance level was set at 5%. All analyses were done using SPSS 26. Cases missing data elements were excluded from analyses.

| RESULTS
After NIS weighing, we identified 106 462 AF ablations performed from 2005 to 2013 who met the inclusion criteria. Table 1 describes the demographics of all patients categorized by obesity and diabetes.
The mean age of the overall patient population was 62.8 years         Tables 2 and 3 describes the annual rates of various grouped complications in obese versus non-obese (Table 2) and diabetic versus non-diabetic patients ( We also sub-stratified our patient population by obesity and dia-  Lastly, Figure S3 shows Catheter ablation is becoming standard-of-care for many AF patients due to its association with improved outcomes and qualityof-life 25 compared to medical therapy. However, it does not come without risks. It is performed while patients are taking oral anticoagulation, and most are still lengthy procedures performed under general anesthesia. 26 Advanced techniques such as ultrasound guidance for vascular access and trans-septal puncture have made the procedure safer in patient with comorbidities such as obesity. 27,28 Nevertheless, these advanced tools are not widely used in lowvolume community centers. NIS real-world data in our study shows that the number of ablations performed on patients with obesity, diabetes or both has increased over the years, compared to those without both comorbidities, and most were performed in low-volume centers. Clinical trials and studies from specialized centers have shown that obesity and diabetes were not significantly associated with increased complication rates. 16,17,24 However, most of these studies involved experienced operators at high-volume centers. One NIS study from 2000 to 2010 showed an estimated procedural complication rate of 6.29%, 23 compared to 2.9% reported in a meta-analysis. 29 In another study using the national readmission database, 1 in 200 patients died within 30 days and mortality was independently linked to procedural complications. 30 Although mortality rates in this study were probably inflated due to selection bias prompted by the We also demonstrate that in all captured AF ablations, obesity was independently associated with increased complication rates (driven primarily by vascular/hemorrhagic and respiratory complications), longer LOS and higher charges after adjusting for demographics, clinical variables, and hospital volume and size. Increased vascular/hemorrhagic complications in obese patients may be due to difficulty in attaining vascular or trans-septal access even with ultrasound guidance and inadequate hemostasis after catheter removal.
Moreover, obese patients undergoing general anesthesia or conscious sedation during this relatively lengthy procedure may experience higher rates of respiratory complications related to worse respiratory mechanics. 35 Obese patients have been shown to have significantly longer procedural duration than non-obese patients in a metaanalysis, 18  The literature extensively describes that poorly controlled obesity and diabetes lead to increased AF recurrence rates after catheter ablation. [13][14][15] Risk factor modification like weight loss with or without bariatric surgery, glycemic control, blood pressure control, and sleep apnea treatment have been shown to increase arrhythmia-free survival rates. 36 We hypothesize that risk factor modification through weight loss prior to AF ablation may also reduce procedural complication rates and charges, but further prospective evaluation is ultimately required. While novel ablation techniques such as pulsed field ablation promise shorter procedures, 37

| LIMITATIONS
The NIS is a de-identified administrative database making it difficult to validate individual ICD-9-CM codes. This significantly affects the sensitivity and specificity when applying the diagnostic codes. Studies based on data mining are susceptible to errors related to coding. Obesity and diabetes were defined using ICD-9 codes, which may not be reliable particularly prior to electronic health records. In addition, outcomes related to the severity of obesity were not examined due to insufficient BMI coding in our sample (<10%) and the potential for selection bias. The data from the NIS lacks the level of detail and patient phenotyping available from clinical trials and registries. As previously described, our study is subject to selection bias towards AF ablations requiring admission due to the two-midnight rule. However, censoring our analysis at 2013 was performed to minimize this bias.
This limitation is also offset by the larger sample size from the NIS and the absence of reporting bias introduced by selective publication of results from specialized centers. This unfortunately also means we

| CONCLUSION
Our study involving real-world data from the NIS demonstrates increased annual complication rates in patients with obesity or diabetes who underwent inpatient AF ablation from 2005 to 2013. Obesity, but not diabetes, is an independent predictor for higher complication rates, longer LOS and higher hospital charges after adjusting for demographics, clinical variables and hospital volume and size. Granted, our database contains only patients who underwent AF ablation as inpatients from the start or were converted to inpatient after developing complications. Weight loss before ablation may help reduce procedural complications and costs.