Association between body mass index and mortality in gram-negative bloodstream infections

Background:

Obesity is an established risk factor for many infection types. Emerging data demonstrate increased body mass index (BMI) may paradoxically protect against mortality in several chronic disease processes. The interaction between BMI and mortality in gram-negative bloodstream infections (GN BSI) has not been defined. This retrospective cohort study examines the impact of BMI on mortality in patients with GN BSI.

Methods:

Hospitalized adults with first episodes of GN BSI from January 1, 2010 to June 30, 2015 at Palmetto Health Hospitals in Columbia, SC, USA were identified. Patients were grouped by BMI in kg/m² as normal weight (18.5 – 24.9; reference), overweight (25.0 – 29.9), obesity class I or II (30.0 – 39.9) and obesity class III (≥ 40). Multivariate Cox proportional hazards regression was used to examine risk factors for 28-day mortality following GN BSI.

Results:

Among 1,069 unique patients with GN BSI, 352 (33%) were normal weight, 319 (30%) overweight, 303 (28%) obese class I/II and 95 (9%) obese class III or morbidly obese. Overall, median age was 65 years and 54% (575) were women. Escherichia coli was the most common pathogen (545, 51%) and the urinary tract was the most common source of infection (545, 51%). Independent risk factors for mortality following GN BSI included age (adjusted hazards ratio [aHR] 1.46 per decade, 95% confidence intervals [CI] 1.28-1.65), cancer (aHR 4.10, 95% CI 2.28-5.96), liver cirrhosis (aHR 3.41, 95% CI 1.84-5.88), non-urinary source (aHR 1.58, 95% CI 1.09-2.23), Pitt bacteremia score (aHR 1.54 per point, 95% CI 1.46-1.62) and inappropriate empiric antibiotic therapy (aHR 2.44, 95% CI 1.47-3.84). Overweight body size was independently associated with reduced mortality (aHR 0.63, 95% [CI 0.40 – 0.96). Obesity class I/II (aHR 0.93, 95% CI 0.60 – 1.44), obesity class III (aHR 0.97, 95% CI 0.42 – 1.98), and diabetes mellitus (aHR 1.05, 95% CI 0.71 – 1.53) were not associated with mortality.

Conclusion: Compared to those with a normal BMI, overweight patients had improved survival following GN BSI. However, this potential protective effect was lost in obese and morbidly obese patients. Further investigation into adequacy of antibiotic dosing in this population is warranted.