Research statistics guidelines

[RCB]

Al,
In a recent discusssion in the Valve section under the thread "Valve question", post #102 the subject of warfarin came up(I know-why does it get talked about so much over there and not here!). This study quoted several parameter relating to valve perfromance in terms of warfarin affectiveness.
One statistic that was discussed was "bleed rate". I had asked you
about this before and you give me a diffinition. It seem to be completely different then the one quoted in this study.
My question to you is, can you take a look at the thread and comment.
My concern is, I thought that Gurkenmier et al., had statitical standards for
warfarin reseach about the compilation of adverse effect from warfarin. Therefore when one research team does a trial of warfarin annd reports a bleed rate, etc., one could be sure that it was the same standard for another study so a comparison could be made. If this is not the case, then it seems to me no comparison could scientifically be made. In other words, they are all
trying to play the same music, but they are all out of key.
Below are the quotes you will find at the above location. I look forward as always to your expert advice. It may be more informative if you post your answer on the other thread.


What is Valve-Related Morbidity?
-It simply means that the person has active valve disease. Not surprising in people who started out with it ten or twenty years earlier.

What do you think they mean by a "bleeding event"?
- It means a bleeding event significant enough that the person sought treatment for it. It doesn't imply life-threatening.

 

[RCB]

Ah- a little help AL

Ah- a little help AL

Is there something about this question that you have no comment?

Al,
In a recent discusssion in the Valve section under the thread "Valve question", post #102 the subject of warfarin came up(I know-why does it get talked about so much over there and not here!). This study quoted several parameter relating to valve perfromance in terms of warfarin affectiveness.
One statistic that was discussed was "bleed rate". I had asked you
about this before and you give me a diffinition. It seem to be completely different then the one quoted in this study.
My question to you is, can you take a look at the thread and comment.
My concern is, I thought that Gurkenmier et al., had statitical standards for
warfarin reseach about the compilation of adverse effect from warfarin. Therefore when one research team does a trial of warfarin annd reports a bleed rate, etc., one could be sure that it was the same standard for another study so a comparison could be made. If this is not the case, then it seems to me no comparison could scientifically be made. In other words, they are all
trying to play the same music, but they are all out of key.
Below are the quotes you will find at the above location. I look forward as always to your expert advice. It may be more informative if you post your answer on the other thread.


What is Valve-Related Morbidity?
-It simply means that the person has active valve disease. Not surprising in people who started out with it ten or twenty years earlier.

What do you think they mean by a "bleeding event"?
- It means a bleeding event significant enough that the person sought treatment for it. It doesn't imply life-threatening.

 

[tobagotwo]

If this is associated with the definition of "bleeding event" in the study posted earlier, there are a couple of thoughts that might fit in and make sense, although they might not address the overall issue in the manner that you wish...

- If there are standards for bleeding events, they are probably fairly recent. They may postdate the study.

- When doing studies, right or wrong, investigators generally get to define their own terms. As long as they do so adequately, so the study can be interpreted reasonably, it seems to be tolerated. Sort of like the differences between legends on maps.

- The likelihood is that using data that goes back twenty years, the criterion for seriousness that is easiest to verify is whether medical help was sought for the wound. It is not likely that there would be accurate records of how much blood was actually lost during an event, or the rate of loss, since some would obviously be lost before medical attention was received.

- I'm not certain there is even anything in place now to routinely and accurately measure blood loss from a bleeding event that commences off of hospital grounds.

Best wishes,

 

[Guest]

I agree that everyone sets their own definitions. It does make comparisons difficult. I had never heard that there was a generally agreed upon set of standards.

My definition of a minor bleed is one that causes a person to stop what they are doing and take action on it. It goes all the way up to just short of requiring 2 units of blood.

A major bleed is one that requires 2 units of blood or more or any bleeding in the head.

 

[RCB]

Thanks for the Post Bob

Thanks for the Post Bob

I guess I am really confused about this one.
When I helped my wife on her dissertation, the AMA/APA
guidelines called for commonly defined uses of terms in the Standards
and Methods section. Since her research was using pts. from a much larger national study that her principle advisors was doing, all her protocols could
not be in conflict with the large study. When she came across a some studies that suggest a problem with the large studies, there was hugh discussion about generally accepted terms when used in research. Having had this experience and the problems that arise whenever a study is free
to decide what the definition is of commonly used terms that professional
use in research and when the interact with lay people here is my problem:
Since in most cases, when discussing the adverse effects that mech valves have WITH warfarin therapy(generally qouted in he 1-4 % range), how with accuracy, can they quote a statistic to a pt. with out also giving a long
explanation of what the standards were used for that statistic.
For the example, your cardiologist tells you the bleed rate on warfarin is 2% /100 pt. yrs., that seems clear enought. The problem arises when there
is no standard definition of "bleed rate". What if one study defines it as having sought treatment at a medical facility. Suppose another defines it as
the need for replacement of at least one pt. of blood. Can you see the problem with this approach to defining research terms. I can guarantee that the former case will produce a much higher number(percentage wise) than the latter case. The former being more based on a qualitative judge- some people are quicker to seek treatment from an ER than other and the latter more quanitative- a doctor can measure both the amount and blood needed.
So how many pts. ask their doctor to define terms when the doctor gives them a statistic. If asked, how many doctors do you think could recall the definition of the term as it was defined by a study. In the case of the visit to an ER visit it could be 4-6%, while in the higher standard of the measured
pt. of blood that needs to be replaced, it could be as low as 1%. Now it doesn't take a PhD. in math to know that 400% to 600% difference is statistically significant.
This doesn't even take into account the differences in warfarin management( or mismanagement as many members here will tell you) between
how warfarin is managed today compared with 30, 20, 10 or even five years ago. In other words, long term studies become worthless because warfarin managment is so dynamic. I think the newer the study(and shorter), the better the data will reflect the state of warfarin management today.
I don't know if the data shows this, but personally my warfarin management is head and shoulders above what it used to be even a year ago, now that I am home testing. I doubt that is reflected in doctors who are now quoting statistic on warfarin use to their pts. considering valve surgery. I would be curious if your doctor even mention it as a parameter in your choice of valves. I know no one mention it to me, till I found this site.

 

[RCB]

Difficult?

Difficult?

I agree that everyone sets their own definitions. It does make comparisons difficult.

I would say it makes it IMPOSSIBLE!

Thanks for the reply- if you say it is true- It is true!

 

[Guest]

Your wife had preceptors to keep her in line. Professors don't have preceptors.

It is like the "Golden Rule" - the guy with the gold makes the rules.

I can't recall ever seeing two studies that defined everything exactly the same.

 

[RCB]

I learn a lot of things here.

I learn a lot of things here.

I just wish my wife could have had you as her professors. Our lives would be a lot better for it.

Your wife had preceptors to keep her in line. Professors don't have preceptors.

It is like the "Golden Rule" - the guy with the gold makes the rules.

I can't recall ever seeing two studies that defined everything exactly the same.

 

[tobagotwo]

I agree, RCB. The sliding definitions provide difficulty in understanding and comparing the results of studies, and even some nonregulated clinical data.

It's one of the reasons I have become such an obnoxious pain in the behind about interpretation of study results. I find it difficult to go through these materials without "skipping ahead," with a preformed idea of what the results are going to represent. I assume that happens to other people, too. Unfortunately, it's usually buried in the thick, tedious verbiage that precedes the numbers and graphs I come looking for. It always seems to take more time than it should to come to an understanding of a study's terms and precepts.

This is part of what makes apples-to-apples comparisons of study reports so very difficult. It usually takes more prose to explain the ways in which the compared results are not level-field than it does to display the results themselves.

Best wishes,

 

[Blanche]

Indeed there are problems in comparing items that use the same term but not the same definition. In order to get around problems in terms, in many studies you will find the term "operational definition," which merely means, this is how the term is being defined here and for purposes of this study. In reading educational research materials, I have found that operational definitions are commonly used. This does clarify a study at hand but certainly does nothing to assist with the problem of comparison.
Regards,
Blanche

 

[Guest]

This is not about warfarin, but it gives you an idea of the "state of the art" concerning the dosing of medications. It was published in today's British Medical Journal. They compared the four "best" sources in the world for advice on how to reduce doses of medications when the person has reduced kidney function. This was their conclusion:

The remarkable variation in definitions and
recommendations, along with scarce details of the methods
used to reach this advice, makes the available sources of drug
information ill suited for clinical use. The methods used to
retrieve information and use data should be described and
made available to the reader. Advice on drug prescription, dose
and dosing interval, contraindications, and adverse effects
should be evidence based.