心不全患者におけるアルゴリズム生成薬物最適化スコアと臨床転帰との関連

Guideline-directed medical therapy (GDMT) implementation in heart failure with reduced ejection fraction (HFrEF) remains suboptimal. A computable algorithm was developed to generate a medication optimization score (MOS) and provide guideline-based recommendations. This computable algorithm was previously validated using clinical trial data, and an updated version was developed in 2021 to include sodium-glucose co-transporter 2 inhibitors. This study evaluated the association between the medication optimization information generated by this version of the algorithm and clinical outcomes using real-world data.

We conducted a retrospective cohort study of 1352 ambulatory adult patients with chronic HFrEF who received care from the advanced heart failure service at the University of Michigan between July 1, 2021, and October 14, 2024. The algorithm-generated MOS was calculated using electronic health record data. The primary outcome was a composite of all-cause mortality or hospitalization. Cox proportional hazards models were used to evaluate the association between baseline MOS and the primary outcome. A time-varying Cox model using the running cumulative MOS and a marginal structural model (MSM) was also conducted. A linear mixed-effects model was used to assess improvement in MOS over time as the secondary outcome.

In the analysis adjusted for HF severity and comorbidities, baseline MOS was associated with a lower hazard of the composite outcome (hazard ratio (HR) 0.96, 95% confidence interval (95% CI): 0.92, 0.99, p = 0.040). In the cumulative time-varying Cox model and the marginal structural model, the association with time-varying MOS became stronger, with HRs of 0.88 (95% CI 0.81-0.95; p = 0.0015) and 0.88 (95% CI 0.83-0.93; p < 0.001), respectively. The event rates per 100 person-years were 44.1 in MOS 0%-33%, 39.5 in MOS 34%-66%, and 31.8 in MOS 67%-100%. Longitudinally, MOS improved over time.

Higher algorithm-generated MOS values were significantly associated with lower all-cause mortality or hospitalization, and the MOS values increased over the follow-up period. This suggested that this algorithm effectively identifies opportunities for GDMT optimization in real-world clinical settings.