(1380) A Novel Propensity-Score-Weighting Method for RAAS Versus Non-RAAS Therapy in Septic Shock Patients

Introduction: Challenges arise when comparing therapeutic effects on patient outcomes retrospectively, as different combinations of therapies sometimes imply higher acuity and risk of death. For example, the addition of Angiotensin II (Ang II) to catecholamines and vasopressin (Conventional Therapy) typically signifies higher acuity, rendering it difficult to know whether Ang II is associated with improved outcomes. We sought to use a novel propensity-score weighting process to balance patient acuity in order to compare the incremental effect of Ang II on patients who are already receiving Conventional Therapy.

Methods: We performed a retrospective chart review of 1149 patients who received vasopressors for septic shock, including 852 who received Conventional Therapy and 297 who also received Ang II. Sequential organ failure assessment (SOFA) score, serum lactate, and Charlson Comorbidity Index (CCI) differed between groups (all p-values < 0.001). Propensity scores were calculated as the conditional probability of assigning a specific treatment considering observed baseline covariates, and promoting similar distributions between the groups. We applied inverse probability of treatment weighting (IPTW), calculating a weighted sample and corresponding weighted analyses, to achieve covariate balance. To reduce potential extreme weights, we introduced a stabilized weighting (SW) method of estimating average treatment effect (ATE), which preserved the sample size. We assessed the absolute standardized differences (ASD) before and after calculating ATE_SW to ensure that the distribution of relevant covariates was similar among groups.

Results: After applying ATE_SW weighting, age, gender, SOFA, CCI, and serum lactate all achieved an ASD < 0.1, indicating sufficient covariate balance. After applying IPTW, the odds ratio of 30-day mortality in the Ang II cohort dropped from 1.35 (95% CI 1.03-1.77, p=0.030) to 1.07 (95% CI: 0.82-1.40, p=0.609).

Conclusions: IPTW may be appropriate for comparison of different therapeutic options when such options imply higher acuity and risk of poor outcomes. IPTW achieves a balanced level of acuity of illness, and ATE_SW weighting is less impacted by extreme values in the data, allowing for preservation of sample size.