Better survival with mechanical vs bioprosthetic AVR

[Silver Bullet]

A new paper from November shows better survival with mechanical compared to bioprosthetic AVR. It's a well conducted observational study from Sweden, though it's not the best quality evidence, which would be a randomized controlled trial. There were significant differences between the groups (the mechanical valve group was younger and healthier) but the investigators use sophisticated statistical analyses to adjust as well as they could for those baseline differences in their analysis.

The improved survival was only seen in pts in their 50s and did not extend to pts over 60.

Consistent with other studies, including randomized trials, stroke rates were the same, but reapportion risk was higher with a bioprosthetic valve. Also interesting were the very low rates of major bleeding in the mechanical AVR group. Sweden is known for having outstanding anticoagulation clinics.

Here's the reference:
Glaser N, Jackson V, Holzmann MJ, et al. Aortic valve replacement with mechanical vs biological prostheses in patients aged 50–69 years. Eur Heart J 2015; DOI:10.1093/eurheartj/ehv580.

There is an editorial on it in the same issue which is excellent if you can get your hands on it.

REMEMBER: this is NOT a randomized trial.

 

[Silver Bullet]

Here's the conclusion of the editorial: "In conclusion, the study of Glaser et al. contributes significantly to
the existing literature on mechanical vs. bioprosthetic valves by
showing that patients above the age of 60 years may benefit from
receiving a bioprosthetic valve, but supporting the notion that lowering
the age for implantation of bioprosthetic valves to below 60
years is premature. New advancements may reduce the risk of adverse
events and potentially mortality after both mechanical and
bioprosthetic valve implantation. However, before significantly altering
treatment strategies, an evidence base from randomized clinical
trials is essential. Although recruitment in a trial comparing
mechanical and bioprosthetic valves would be cumbersome because
of strong patient preferences in favour of bioprosthetic valves,
it is our duty as individual physicians as well as Heart Team members
to provide full information to patients regarding their long-term
risk–benefit ratio, which should include evidence of a survival benefit
of mechanical valves in patients below 60 years of age."

 

[pellicle]

Hi
Interesting finding, matches some other results I've seen.

You may find this interesting

http://mayo.img.entriq.net/htm/MayoPlayer1.html?articleID=4071

I would question if a random trial is somehow "better". The primary benefit lf randomized trials ks to maximize" transferability to a random member lf the public. Matching on criteria while more specific actually gives better matches jf you match the study criteria.

Thanks for posting.

 

[Silver Bullet]

The primary advantage of randomized trials is not generalizability to the population. Real world observational data has that advantage because the subjects are the general population of interest.

The primary advantage of RCTs is that the 2 groups should be equal (which is what the sample size and randomization process should achieve) except for the treatment they get, permitting causal conclusions to be drawn about the treatment specifically. Because RCTs have inclusion and exclusion criteria up front, limiting participation in the trial to a subset of the general population of interest, they are often criticized for not being generalizable.

The best quality evidence for a therapy is widely considered to be multiple adequately sized RCTs, or systematic reviews (meta-analyses) of multiple adequately sized RCTs.

 

[pellicle]

so I guess that its appropriate to suggest that a 30year old will have the same propensity (risk) for bleed episodes when on warfarin? That is essentially what fully randomised trials do. I would argue that you can not expect to see the same number of bleed events on a healthy 30yo who is on warfarin as a 80yo ... yet that is exactly what is implied in interpretation of studies. Without appropriate matching of cohorts and criteria the information loses information.

RCT is not quite the same as a randomised trial, I can't see how a control group is created in a valve study as they would have to undergo no treatment.

 

[Northernlights]

In a randomised controlled trial the control group can either be assigned to a placebo, or an established treatment. Both types are randomised controlled trials. It's the method that counts, not what it's controlled against.

Randomised controlled trials are indeed the gold standard, as otherwise it is very difficult not to introduce bias. Paired propensity score matching (as in the above trial, which matched 1099 pairs of patients from an initial cohort of 4545 patients) tries to get round this by matching individual pairs of patients, one from each group, to minimise bias. It's far from ideal though. But for a randomised trial of mechanical v. bioprosthetic valves it would be probably impossible to recruit enough patients, as the editorial says, so it's unlikely to happen.

 

[pellicle]

Randomised controlled trials are indeed the gold standard

Agreed ...

as otherwise it is very difficult not to introduce bias

Agreed

. Paired propensity score matching (as in the above trial, which matched 1099 pairs of patients from an initial cohort of 4545 patients) tries to get round this by matching individual pairs of patients, one from each group, to minimise bias. It's far from ideal though. But for a randomised trial of mechanical v. bioprosthetic valves it would be probably impossible to recruit enough patients, as the editorial says, so it's unlikely to happen.

That was the point I was hedging at. I did a bit of study in research design during my masters (by resarch , not coursework (spoon fed) thesis preparation. The stuff done so far is all good work, but next step in value adding to (or extending the findings of) existing research knowledge, is to go the extra yards and stratify more carefully. As you probably know the aim is to test transferrability of findings. If the population under study is significantly different based on a significant parameter then the research findings may not be applicable.

My contention in much of this research (related to valves) that we compare 70year olds with 25 year olds as if they were a homogenous group, and there are significant differences such to warrant more specif age related research.

I suspect we are on the same page here and its just my initial choice of words and the context free nature of text discussions with a massive variance in authors that makes it seem like we are saying different things.

Best wishes