Methods: We harvested the MIMIC-III database, a deidentified publically available dataset of 59,000 intensive care unit admissions, for patients with a confirmed diagnosis of sepsis and medication lists. Patients with immunosuppressive medications or conditions were excluded. Data were normalized and adjusted for Elixhauser comorbidity index components after propensity score matching. Organ dysfunction was analyzed by AST/ALT, estimated glomerular filtration/creatinine clearance (eGFR/CrCl), and troponin I/T peak or nadir, respectively, as well as SOFA score and peak lactate. We also analyzed 28-day and in-hospital mortality.
Results: We matched n=6,224 statin users and non-users. Mean age was 72 years and 54.5% of patients were female; 31% and 12% of patients had uncomplicated and complicated diabetes, respectively; and 4% were alcohol-dependent. Mean AST/ALT were 88 vs 90 and 72 vs 75 U/l (p>0.05 for both); eGFR/CrCl were 60 vs 63 and 66 vs 67 ml/min/1.73 m2 or ml/min (p=0.001 and p>0.05, respectively); troponin I/T were 8.3 vs 5.7 and 0.56 vs 0.40 ng/ml (p>0.05 and p=0.001, respectively). SOFA score was 5.45 vs 5.68 (p=0.005) and peak lactate 1.9 vs 2.0 mmol/L (p=0.03). 28-day and in-hospital mortality were 18.8 vs 22.7% and 16.0 vs 20.5% (p<0.001 for both).
Conclusions: Statin use prior to sepsis was associated with lower short-term mortality that was clinically and statistically significant. However, this mortality benefit seems not explained by similar markers of organ dysfunction and only marginally improved SOFA scores and peak lactate. Further research is necessary to elucidate if decreased inflammatory response without decreased organ dysfunction is attributable.