Number needed to treat - Interesting Tid bits

[Aloha Kid]

Just heard an interesting speaker Dr David Newman - Research EM Columbia U. Here are some of the numbers he cited in his lecture today regarding numbers needed to treat. Or in other words, how many patients I have to treat with (SAID) therapy in order to save a life or whatever intended goal you might have.


Rapid defib of V-fib NNT 3
Hypothermia for cardiac arrest NNT 6
Sepsis Early Goal Directed Tx NNT 6
BIPAP for COPD NNT 10
Low Tidal Volumes/ARDS NNT 12
ASA for MI NNT 40
PCI NNT 30-40
ASA (post MI) prevention NNT 100
Statin (known CAD) prevention NNT 400
Statin (No CAD) prevention NNT 5000
Seat belt for MVA NNT 25,000


BBlocker for STEMI NNT - Infiniti
Mammograms - death prevent NNT - Infiniti
GIIbIIIa for STEMI NNT - Infiniti
GIIbIIIa for NSTEMI NNT - 250 (only a 30day benefit)

ABX for COPD (preventing bounce back) NNT - 3
Prednisone for Asthma (as above) NNT - 10
ABX for COPD (as above) NNT - 12

---

Members don't see this ad.

 

[WilcoWorld]

Very interesting perspective!

Could you explain the "infinity" calculations in detail (is this just a consequence of CI's crossing 1)?

 

[EvoDevo]

Just heard an interesting speaker Dr David Newman - Research EM Columbia U. Here are some of the numbers he cited in his lecture today regarding numbers needed to treat. Or in other words, how many patients I have to treat with (SAID) therapy in order to save a life or whatever intended goal you might have.


Rapid defib of V-fib NNT 3
Hypothermia for cardiac arrest NNT 6
Sepsis Early Goal Directed Tx NNT 6
BIPAP for COPD NNT 10
Low Tidal Volumes/ARDS NNT 12
ASA for MI NNT 40
PCI NNT 30-40
ASA (post MI) prevention NNT 100
Statin (known CAD) prevention NNT 400
Statin (No CAD) prevention NNT 5000
Seat belt for MVA NNT 25,000


BBlocker for STEMI NNT - Infiniti
Mammograms - death prevent NNT - Infiniti
GIIbIIIa for STEMI NNT - Infiniti
GIIbIIIa for NSTEMI NNT - 250 (only a 30day benefit)

ABX for COPD (preventing bounce back) NNT - 3
Prednisone for Asthma (as above) NNT - 10
ABX for COPD (as above) NNT - 12

Great post! I wonder if there is some sort of link where you can look this sort of thing up?

 

[BKN]

Great post! I wonder if there is some sort of link where you can look this sort of thing up?

you can do your own, it's quite easy. The formula is NNT = 1/(CER-EER).

CER (Control efficacy rate) stands for the major outcome rate in the control group. EER is the rate in the experimental group. the term (CER - EER) is also, known as the absolute risk reduction (ARR).

Be careful about using lists of NNT or any other summary measure. You don't know what the efficacy measure was, nor the population tested, nor whether randomized, . . . and all the other quality measures for a experimental study.

For example, Rapid defib of V-fib NNT 3 looks very good indeed. But what's the definition of rapid? Is it within five minutes of collapse, or two minutes of defibrillator arrival or what? what's the definition of success? Is it discharge from the hospital without neurologic damage? I doubt it very much. Seems much more likely, the definition was return of systemic circulation. A long ways from success.

My point is, you must look at the original study to know what the summary measure means.

Cheers,
BKN

 

[EvoDevo]

you can do your own, it's quite easy. The formula is NNT = 1/(CER-EER).

CER (Control efficacy rate) stands for the major outcome rate in the control group. EER is the rate in the experimental group. the term (CER - EER) is also, known as the absolute risk reduction (ARR).

Be careful about using lists of NNT or any other summary measure. You don't know what the efficacy measure was, nor the population tested, nor whether randomized, . . . and all the other quality measures for a experimental study.

For example, Rapid defib of V-fib NNT 3 looks very good indeed. But what's the definition of rapid? Is it within five minutes of collapse, or two minutes of defibrillator arrival or what? what's the definition of success? Is it discharge from the hospital without neurologic damage? I doubt it very much. Seems much more likely, the definition was return of systemic circulation. A long ways from success.

My point is, you must look at the original study to know what the summary measure means.

Cheers,
BKN

Thanks for the teaching point. 🙂 I appreciate the reminder of the basis for NNT. 👍

I hear what you're saying w/r/t being aware of summary numbers. I guess what I meant to say, has Annals, SAEM, the NIH, Cochrane reviews or some neutral third party perhaps come up with NNT summary table that they consider valid? You know, how Annals at the end of the year has the big summary issue where they rate various categories of treatment in the ED?

 

[mlw47]

This is worth a read:


The "number needed to treat" turns 20 — and continues to be used and misused

[FONT=verdana,arial,helvetica] Finlay A. McAlister, MD MSc

.CMAJ • September 9, 2008; 179 (6). doi:10.1503/cmaj.080484.

http://www.cmaj.ca/cgi/content/full/179/6/549

 

[southerndoc]

I have a lot of concern with the NNT for seat belts being 25,000. It's misleading because the major benefit of seat belts is morbidity reduction.

 

[jbar]

I have a lot of concern with the NNT for seat belts being 25,000. It's misleading because the major benefit of seat belts is morbidity reduction.

Also you need a time scale, is that 25,000 drivers per mile, per day, lifetime etc? Over a lifetime just about everyone is in an accident, so I'd expect the NNT to reduce injury level would be closer to 2 or 3 over the course of a life.

 

[nymbarra]

I believe there is also a number needed to harm, ie those given the therapy who experience adverse effects. A good example is bleeding in Xigris for septic shock. Both NNH and NNT should be used when deciding on treatment.

http://www.medicine.ox.ac.uk/bandolier/booth/glossary/NNH.html

 

[Aloha Kid]

you can do your own, it's quite easy. The formula is NNT = 1/(CER-EER).

For example, Rapid defib of V-fib NNT 3 looks very good indeed. But what's the definition of rapid? Is it within five minutes of collapse, or two minutes of defibrillator arrival or what? what's the definition of success? Is it discharge from the hospital without neurologic damage? I doubt it very much. Seems much more likely, the definition was return of systemic circulation. A long ways from success.

Cheers,
BKN

The study for rapid defib was based out of Las Vegas Casinos. As soon as someone went down, security (trained with AED usage) ran to the person applied the pads and pressed the button.

What is interesting is as you know, casinos have cameras everywhere. As soon as a person hits the ground someone knows about it. There was apparently actual footage of a guy throwing dice at a table. He goes down. Security moves in and shocks him. He gets up like nothing happened and picks up the dice wanting to keep playing while security and EMS try to convince him to go to the hospital.

This is the definition of rapid defibrillation.

 

[Apollyon]

What is interesting is as you know, casinos have cameras everywhere. As soon as a person hits the ground someone knows about it. There was apparently actual footage of a guy throwing dice at a table. He goes down. Security moves in and shocks him. He gets up like nothing happened and picks up the dice wanting to keep playing while security and EMS try to convince him to go to the hospital.

A friend of mine heads EMS at a casino. They're 11/15 on saves by defib, and, just like you say, you can follow the response to the step, camera by camera.

 

[BKN]

This is worth a read:


The "number needed to treat" turns 20 — and continues to be used and misused

[FONT=verdana,arial,helvetica] Finlay A. McAlister, MD MSc

.CMAJ • September 9, 2008; 179 (6). doi:10.1503/cmaj.080484.

http://www.cmaj.ca/cgi/content/full/179/6/549

Thanks for this. Best EBM article I've read in a while. The Canadians are still the best at this.

Evo, read this article and you'll understand why you can't "make up an approved list".

I agree with McAlister and the sources he quotes. You've got to go back and read the article to understand the context for each NNT.

BN

 

[roja]

Okay, now that I have stopped belly laughing at seeing Dr. Newman's name on SDN, I will add my few bits to NNT.


NNT can also be easily calculated as NNT= 1/ARR (ARR= absolute risk reduction)

NNT only have meaning when you are talking about treating the people who actually HAVE the disease. A fine point but keep in mind, we often think about treating people without a definative diagnosis. So, when you are calculating NNT keep in mind you are calculating it only on people who have the disease (or should be). IE, if you are initiating EGDT, you should only be caculating the NNT on the patients who are actually septic.

HOWEVER, the NNH is calculated on all patients recieving the treatment, regardless of disease.

🙂



And the infinity means that the CI (usually from the ARR) is not useful. Calculating the NNT when something is not statistically significant CAN be done but turns into a meaningless number: you have to treat an infinite number of people to see a result.

(go back to the original NINDS trial and calculate out a few NNT. you will find that many of thme will be 'infiniti'.)

 

[RxnMan]

...Evo, read this article and you'll understand why you can't "make up an approved list"...

But you could make an approved list using a list of the comorbidities of the study population (the NNT for this drug was 10 in this population of people who smoked > 20pk-yrs, FEV1 < 40%...) or a sub-listing of NNT based on risks (NNT = 100 for smoked >10 pk-yrs, NNT = 50 >20 pk-yrs, and NNT = 10 for ...).

It would become more complex for multi-drug regimens (e.g. ASA / beta blocker / ACE).

I also like the article's discussion on NNT that are not statistically significant.

 

[leviathan]

The study for rapid defib was based out of Las Vegas Casinos. As soon as someone went down, security (trained with AED usage) ran to the person applied the pads and pressed the button.

What is interesting is as you know, casinos have cameras everywhere. As soon as a person hits the ground someone knows about it. There was apparently actual footage of a guy throwing dice at a table. He goes down. Security moves in and shocks him. He gets up like nothing happened and picks up the dice wanting to keep playing while security and EMS try to convince him to go to the hospital.

This is the definition of rapid defibrillation.

I've seen it happen once where the patient arrested after we arrived. He opened his eyes within seconds of delivering the shock!