SLIDE 1
Stefan Anker, MD PhD Charit Medical School Berlin, Germany. CoI: - - PowerPoint PPT Presentation
Stefan Anker, MD PhD Charit Medical School Berlin, Germany. CoI: - - PowerPoint PPT Presentation
Effect of ivabradine on recurrent hospitalization for worsening heart failure: findings from SHIFT Commentary Stefan Anker, MD PhD Charit Medical School Berlin, Germany. CoI: Servier (speakers fees) Congratulations to the authors The
SLIDE 2
SLIDE 3
The Problem
In a trial using the composite of CV death or HF hospitalization, a patient has a ‘primary endpoint’, if he or she experiences CV death as the first of these two possible events or HF hospitalization. A patient may experience CV death after HF hospitalization and may also experience repeat HF hospitalization. Neither of these subsequent events count in the ‘time-to-first’ event analysis.
Hence, ‘time-to-first’ event analyses do not fully reflect the true burden of HF-REF in a contemporary population.
SLIDE 4
A vast amount of information is ignored
% of all observed events ignored in ‚time-to-first‘ analysis CHARM-Added 51.2 CHARM-Alternative 51.4 EMPHASIS-HF 41.5 SHIFT 43.3 I-PRESERVE 41.1 CHARM-Preserved 46.4
SLIDE 5
A vast amount of information is ignored
SHIFT: 1186 patients had at least one HF hospitalization (which contributed to the ‘time-to-first’ event analysis, along with 544 CV deaths). 472 patients had two or more admissions and there were 2113 hospitalizations for HF and 940 CV deaths in total. This means, 44% of all HF hospitalizations and 42% of CV deaths were ‘ignored’ in the (primary) ‘time-to-first’ event analysis of SHIFT.
SLIDE 6
What was found is an important effect
SLIDE 7
What was found is an important effect
SLIDE 8
What was found is an important effect
‘Time-to-first-event‘ analysis: Ivabradine treatment prevented 47 hospital admissions for HF per 1000 patients treated. ‘Repeat events‘ analysis: Ivabradine treatment prevented 93 hospital admissions for HF per 1000 patients treated. Death prevents hospitalization. These analyses are not adjusted for the impact of death.
SLIDE 9
Analyses not free of complications
Statistical tests:
- negative binomial regression
- Poisson regression (with adjustment for over-dispersion)
- Andersen–Gill method (with robust standard errors)
- method of Wei, Lin & Weissfeld (WLW)
- WIN-ratio
- method of Finklestein & Schoenfeld (F-S)
Power calculation approaches not fully solved. Regulatory acceptance to be resolved. One HF-trial has prospective uses this methodology: CHAMPION. Trials that use this methodology: PHARM-CHF, RESHAPE-HF, one current Novartis programme.
SLIDE 10