Death is not a New York Heart Class

Is data emanating from human transcatheter aortic valve trials showing clinical strength or weakness?  Here is an example – you be the judge.

At the recent TCT conference in San Francisco, the midterm results of the PARTNER EU transcatheter valve study were presented.  This study followed 120 patients implanted with the Edwards Sapien Valve.

Heartwire ran a story on the study results reporting that the patients in the study had dramatic improvements in NYHA classification one year post implant.

Functionally, at one year, 89% of patients had NYHA class 1 or 2 heart failure, with 81% improving by at least one heart-failure class over this period.

I was so impressed with these results that I wanted to look at the actual data.  I went to the TCTMD website and downloaded the 18 mo PARTNER EU results.  Below are the PowerPoint slides showing the basis for the 89% and  81% improvements.

The data looks great upon first inspection  but, after further review, it looks less than great.  Let me summarize the data shown on the charts.

Slide 1 – NYHA Group at One Year

Slide 1 - NYHA Class at Year 1

Slide 1 - Click to Enlarge

  • About 53% of the patients enrolled in the study are in NYHA class I or II
  • About 7 % are in NYHA Class III or V
  • About 38 % are dead (there is no NYHA Class for death)
  • 2 % are lost to followup

Slide 2 – NYHA Improvement at One Year

Click to Enlarge

Slide 2 - Click to Enlarge

  • At one year follow-up, 60 pts of the 118 (2 lost to follow-up) improved at least one NYHA class – this equates to 51% of all patients, not 81% as reported.
  • The remaining 58 pts. saw no improvement or got worse, with 46 being dead by year end.

I summarize this data as follows:

Of the patients that received a Sapien valve, about 50% felt better after one year.  About 80% of the pts. that did not feel better died.  This equates to about a 40%  overall death rate at one year.

Why did the study report 89% in Class I or II and  81% improvement in heart class?  It is because the results are based on analyzing only the patients that survived to year one.  The results do not include the significant number that died and are not in a NY heart class.

Although this might be how the PARTNER study investigators have agreed to analyze the data, it does not seem to be a very fair statistical representation, especially when the death rate is so high.   Using this statistical logic to an extreme, if only one patient survived after one year and this lucky patient also improved one heart class, the dismal study results could be optimistically reported as:

“At year one, 100% of patients improved at least one NYHA class”.

There are other slides in this PARTNER presentation that also eliminate the accumulating dead patients from the success calculations as time moves forward.   Even if this is considered standard practice, isn’t it misleading.  Or is my interpretation wrong?

Maybe it would it be helpful if  death was added as NYHA Class V?   What do you think?

Post a Comment

Your email is never shared. Required fields are marked *