Prior Bariatric Surgery Predicts Lower Life-Threatening Morbidity in Patients Admitted for Acute Allergic Reaction and Anaphylaxis: a Propensity Score-Matched Analysis of the U.S. Nationwide Inpatient Sample, 2005–2018

All data used in this study were obtained through request to the Online HCUP Central Distributor. This study conforms to the NIS data-use agreement with HCUP. The study was exempted from Institutional Review Board (IRB) approval because there was no direct involvement by patients or other persons as the study used secondary data from the NIS database. Because all data in the NIS database are de-identified, the requirement of informed consent was also waived.

This study utilized the International Classification of Diseases, Ninth Revision (ICD-9) and Tenth (ICD-10) diagnostic codes to identify adults ≥ 18 years old with a diagnosis of morbid obesity (ICD-9: 278.01, V85.35-V85.45; ICD-10: E66.01, Z68.4x) and who were admitted to US hospitals with a primary diagnosis of acute allergic reactions/anaphylaxis between 2005 and 2018 as recorded in the NIS database. The patients were further divided into two groups based on whether or not they had prior bariatric surgery (laparoscopic or open Roux-en-Y gastric bypass, laparoscopic adjustable gastric band, and laparoscopic sleeve gastrectomy) through ICD-9 and ICD-10 codes, as shown in previous studies [14]. Patients with no information on in-hospital mortality, discharge destination, hospital costs, and length of stay (LOS) were excluded. The ICD codes for identifying acute allergic reactions/anaphylaxis and bariatric surgery are listed in Supplementary Table 1.

Study endpoints were: incidence of 1) any life-threatening morbidity that occurred during admission; 2) in-hospital mortality; 3) unfavorable discharge, defined as transfer to nursing homes or long-term care facilities; 4) prolonged LOS, defined as LOS >  = 75th percentile; and 5) hospital costs. Significant morbidities included cardiac arrest and dysrhythmia, acute myocardial infarction (AMI) and cerebrovascular accident (CVA), venous thromboembolism (VTE), respiratory distress or failure, pneumonia, and severe infection, sepsis and septic shock, and intubation and mechanical ventilation. The ICD codes for identifying these morbidities are listed in Supplementary Table 1.

All analyses were performed using SAS survey statements to address the complex sampling design for HCUP-NIS data (SAS Institute Inc., Cary, NC, USA). Continuous variables are presented as weighted mean and standard error (SE); categorical variables are presented as unweighted numbers and weighted proportions. Differences in means between the groups were compared using the SURVEYREG statement for continuous variables. Rao-Scott chi-square test was performed to examine differences in proportions between the groups using the SURVEYFREQ statement for categorical variables. Logistic regression was conducted using the SURVEYLOGISTIC statement to determine the associations between study variables and outcomes of interest. To balance the baseline characteristics between the groups (prior bariatric surgery or not), patients in the two groups were propensity score-matched (PSM). Propensity scores assigned to each patient were derived from the multivariable logistic regression model constructed to determine the likelihood of receiving bariatric surgery after controlling variables with a p-value < 0.01 in the unmatched population of the two groups. A 1:4 fixed ratio nearest-neighbor matching was performed. Further, univariate and multivariate logistic regression models were used to calculate the odds ratios (ORs) and 95% confidence intervals (CIs) in estimating the associations between study variables and patient outcomes. Variables that reached statistical significance in univariate analysis were included in multivariate models after adjusting for possible confounding variables. A two-sided P-value of < 0.05 was regarded as statistical significance.

The study sample selection process is depicted in Fig. 1. A total of 21,561 hospitalized patients aged ≧18 diagnosed with morbid obesity and admitted to hospitals for acute allergic reactions and anaphylaxis were identified in the HCUP-NIS database of the US from 2005 to 2018. After excluding patients with missing information on LOS, discharge destination, in-hospital mortality, or hospital costs, 21,126 patients (representing a 105,056 US population) were finally included. Among these, 683 patients had undergone bariatric surgery, and 2,732 patients were not retained for subsequent analyses after 1:4 ratio PSM. Proportions of each primary diagnosis of acute allergic reaction and anaphylaxis are listed in Supplementary Table 2.

Supplementary Table 3 summarizes the characteristics of patients hospitalized for acute allergic reaction and anaphylaxis before PSM. The mean age of the Patients’ mean age was 54.7 ± 0.1 years, and females accounted for 67.5%. Significant differences were found between patients with or without prior bariatric surgery in age, sex, race, income, primary payer, some comorbidities (all p < 0.01), and hospital location. Patients with prior bariatric surgery were younger, with more females, and had fewer comorbidities than those without bariatric surgery. After PSM, the different initial distributions in the characteristics of patients and hospital between the two groups were balanced, as shown in Table 1.

Frequencies of outcomes among patients hospitalized for acute allergic reaction and anaphylaxis are listed in Table 2 and Supplemental Table 4. After matching, patients with prior bariatric surgery had a significantly lower proportion of any life-threatening morbidity (37.2% vs 47.4%), respiratory distress and failure (11.2% vs 17.0%), pneumonia and severe infection (7.4% vs 10.2%), sepsis and septic shock (15.2% vs 20.9%), intubation and mechanical ventilation (11.2% vs 14.6%), prolonged LOS (10.3% vs 20.6%) and unfavorable discharge (6.9% vs 12.5%) than those without prior bariatric surgery, respectively.

A multivariable regression model was constructed to determine the associations between inpatient outcomes and prior bariatric surgery. The results are shown in Table 3. After adjusting for relevant confounders, patients who had prior bariatric surgery were significantly less likely to have life-threatening morbidity during hospitalization than those who had not (aOR:0.64, 95% CI:0.54–0.76). Also, patients with prior bariatric surgery had a significantly lower risk for prolonged LOS (aOR: 0.43, 95% CI:0.33–0.55). However, prior bariatric surgery was not associated with reduced odds of significant mortality or unfavorable discharge. Regarding individual morbidities, significantly lower odds were found in respiratory distress and failure (aOR: 0.61, 95% CI:0.47–0.78), pneumonia and severe infection (aOR: 0.71, 95% CI:0.53–0.96), sepsis and septic shock (aOR: 0.69, 95% CI:0.55–0.85), and intubation and mechanical ventilation (aOR: 0.72, 95% CI:0.56–0.93) in patients with prior bariatric surgery compared to those without prior bariatric surgery.