Which statistic is correct?

[Luisa]

In last weeks" WSJ,there was a small article on "Kicking the Habit". It was written by Sarah Slobin and the source was CDC Office of the Surgeon General; Stephan Hess,James M. McPherson(Lincoln). The small text is as follows: Chances after 6 months a smoker will have quit using these methods alone: Chantix 33.2%,Nasal Spray 26.7%,Patch 26.5%, Gum 26.1%,inhaler 24.8%.

The figures we use are quite different. What are the correct figures and what source can we quote to back up our stats?

 

[Our House]

Not sure about the actual stats themselves (Wikipedia may be a good place to look), but I do know that the standard benchmark is 1 year, not 6 months. It's no surprise that pharmaceutical companies would want to use 6 months in their press releases and advertisements though.

 

[rothenbj]

Even at six months, those numbers appear awfully high. This article may help-

Chantix Helps Smokers Kick Habit

Taking a controlled group of pack a day smokers that smoked about 24 years they showed

The four-week continuous quit rates for the various treatments were:

48% for 1 mg Chantix twice daily
37% for 1 mg Chantix daily
33% for Zyban
17% for the placebo

Then at 1 year Chantix was down to 14% against 5% placebo so you can sell that as the smoker being three times as likely to quit over a cold turkey approach or a smoker has a nine percent better chance of quiting with Chantix than a cold turkey attempt. The second survey they looked at arrived at a one year rate between 19 and 22%, depending on dosage method as opposed to 4% in the placebo group.

Obviously we know nothing about the type smokers that took part in the second study. The other NRT solutions can't possibly be accurate based on my experience and exposure to people that have tried these methods. The real test is at least a year and I'd go so far as looking at 2 and 5 year quit rates. When you consider that the number of smokers hasn't really fallen off much over the last five years, those numbers really look silly IMHO.

 

[keelalagirl55]

Well...their 6 month rates may be correct as I myself have tried to quit 4 different times with medication...well...obviously since it was more than once it did not work, but it did last for at least 4-6 months....lol....I would like to know the start back rate of those that quit after 6 months.....and what do they define as quitting? Does that mean that they did have a significant time where they did not smoke regardless of whether they started back up again or not?

 

[Luisa]

Even at six months, those numbers appear awfully high. This article may help-

Chantix Helps Smokers Kick Habit

Taking a controlled group of pack a day smokers that smoked about 24 years they showed

The four-week continuous quit rates for the various treatments were:

48% for 1 mg Chantix twice daily
37% for 1 mg Chantix daily
33% for Zyban
17% for the placebo

Then at 1 year Chantix was down to 14% against 5% placebo so you can sell that as the smoker being three times as likely to quit over a cold turkey approach or a smoker has a nine percent better chance of quiting with Chantix than a cold turkey attempt. The second survey they looked at arrived at a one year rate between 19 and 22%, depending on dosage method as opposed to 4% in the placebo group.

Obviously we know nothing about the type smokers that took part in the second study. The other NRT solutions can't possibly be accurate based on my experience and exposure to people that have tried these methods. The real test is at least a year and I'd go so far as looking at 2 and 5 year quit rates. When you consider that the number of smokers hasn't really fallen off much over the last five years, those numbers really look silly IMHO.

The numbers I have seen on the ECF have been much much lower. I will have to search for one of our comparison charts,but I seem to remember all the rates were under 10%. I have been quoting that number and I want to make sure that whatever I say is accurate. Thanks for your suggestions.

 

[Turnkeys]

I can't directly comment on Chantix as I've never tried it. When my GP suggested it, I recalled a previous quit attempt with Welbutrin. If I told my GP I was depressed, insurance would cover it. If I wanted the same 'script for a smoking cessation aide, I had to pay out of pocket. So at her wink/nod suggestion, we agreed I was feeling depressed.

Since Chantix doesn't have a similar dual usage the same ruse wasn't an option. It was only later that I learned about the side effects, which I consider unacceptable.

 

[mwa102464]

That's all good, but then you also have to figure in the side effects of these meds like the worst nightmares you will ever have (chantex & the patch) compared to the ecig. I would love to take 100 smokers and put them on a really good ecig than put 100 on Chantix and another 100 on the patch or another med. Then check in a year later and see who the REAL WINNER IS...!!!

 

[mwa102464]

I can't directly comment on Chantix as I've never tried it. When my GP suggested it, I recalled a previous quit attempt with Welbutrin. If I told my GP I was depressed, insurance would cover it. If I wanted the same 'script for a smoking cessation aide, I had to pay out of pocket. So at her wink/nod suggestion, we agreed I was feeling depressed.

Since Chantix doesn't have a similar dual usage the same ruse wasn't an option. It was only later that I learned about the side effects, which I consider unacceptable.

MY POINT EXACTLY, as well as the ecig being the best to use to stop smoking cigs right away and even long term

 

[Turnkeys]

the worst nightmares you will ever have (chantex & the patch) compared to the ecig.

OMG No kidding!!!!

 

[rothenbj]

This study may shine a bit more light-

Nicotine Replacement Therapy Modestly Boosts Long-term Success Rates for Smokers

"Researchers from the Institute of Social and Preventive Medicine in Switzerland identified all randomized placebo-controlled trials of NRT that lasted longer than 12 months. Twelve studies with an average follow-up of 4.3 years and a total of 4,792 participants (NRT: 2,408; placebo: 2,384) met the criteria for this meta-analysis. Three types of NRT were studied: patch (5 trials), gum (4), and nasal spray (3); each delivered about 1 mg of nicotine per hour. Participants used NRT for about three months and received counseling; the control groups only received counseling. The researchers, therefore, were able estimate the long-term effectiveness of NRT beyond the effect of counseling."

"In both the treatment and placebo groups, 30% of volunteers relapsed after the 12-month point. NRT showed significant benefits in six of the 12 trials, and no benefit in the other six. But when all the data were combined and analyzed, NRT almost doubled the rate of successfully quitting beyond 12 months. However, since the rate of success was quite low, even a doubling of this rate meant success for only about 7% of the 2,408 participants receiving NRT in this meta-analysis."

Basically, I read that to mean that the long term effectiveness of cold turkey is around 3.5% with traditional NRT long term success at 3.5% better success, still pretty ineffective. I'm sure Chantix has a better quit rate, but as posted, there are some serious consequences in using the drug. Also working backwards, based on the 30% relapse after one year, the average quit rate at one year would be around 10% for the NRT studied. So if the average of the Spray, Patch and Gum were 26.4% at six months, another 62% were lost between month 6 and one year.

 

[Bustastew]

Vocalek, would you mind posting the stats in NRT and ecig quit rates? I need the most credible source I can get for a pursuasive speech. What about the 2% at 20 months? Wikipidia wont cut it as a credible source =(

 

[Placebo Effect]

You can find the info, and names of the sources, in Elaine's presentation to the FDA. I believe the e-cig quit rate studies can be found on CASAA's site under Resources.

Here's the article that Elaine sources for the 2% cessation at 20 months.

 

[zoiDman]

One of the first things you learn, or should learn, when dealing with statistics is that results can be “adjusted” to about anything you want.

You can easily tweak a test be relaxing the Confidence Interval or the Strength of the Test. If that doesn’t suit you, you can always resample from a sample group of your choosing. When these fail, start throwing out data until the desired result is obtained.

If a statistical result doesn’t include, but not limited to such parameters as…

Who conducted the Test? Possible biases?
Who paid for the Test? Almost Always a Bias!
How the sample was chosen? Ie. Random, Demographic, Cross Sectional
What is the sample size?
Was any data types omitted from the test?
What type of test is used? Ie. Controlled. Blind. Double Blind.
What Confidence Level are the results based on?
Do the results reflex other Independent test results?

… then your not being told anything.

I had a college professor tell me one time that “Statistics is Great Mathematical Tool to Fool the Ignorant.”

 

[Vocalek]

You can find the info, and names of the sources, in Elaine's presentation to the FDA. I believe the e-cig quit rate studies can be found on CASAA's site under Resources.

Here's the article that Elaine sources for the 2% cessation at 20 months.

Placebo Effect has posted the correct source that I used. The key difference is that in Moore's analysis, the results were limited to NRT "when used as directed." I did mention that in my FDA presentation.

The directions are to wean down and totally off nicotine, usually over the course of 12 weeks. It makes perfectly good sense that those people who don't follow the directions and who continue to use the products as long as they have any urges whatsoever to smoke would have higher rates of being quit at various points in time.

And if the powers that be hadn't been so doggone focused on keeping the products from being of interest to kids, there would have been more pleasant flavors available from day one, the nicotine dosages would be much higher (safety/effectiveness testing has been done on patches as high as 42 mg), and the directions would not tell smokers to stop using them. Millions more smokers might have quit if the focus had actually been on the purpose for creating the products in the first place!

 

[rolygate]

It's generally recognised that the best figure for someone who is not a non-medical researcher of stature to quote is a 7% success rate for NRTs at 20 months, as this is apparently easily provable from published work; while if you are in the trade and therefore likely to be able to call on help from colleagues in the event of being on the wrong end of a civil suit from a pharma manufacturer, you can quote 2% (as Dr Nitzkin of the AAPHP does). In other words if you are an ordinary Joe then docs show that NRTs have a long-term failure rate of 93%, but if you are a name then you can tell the truth and say it's 98%.

Apparently the real efficacy of NRTs is only a fraction better than a placebo, if you are measuring the long-term benefit - but it might be wise to be in the trade if you want to go around stating that as fact.

Pharma would be able to crush anyone except eminent doctors who can line the court with their pals. So, the non-medical writers we know who have to produce a figure tend to quote a 7% success rate, as apparently that is easily provable. But here is a link to Dr Nitzkin's work, he is not afraid to quote a 2% success rate for NRTs:

http://www.e-cigarette-forum.com/forum/pdf/ecigcontext.pdf
[2% NRT success rate quoted at para #8 and ref #5]

A quick survey by Dr Siegel indicates that ecigs are about 3 times more effective than NRTs at the short term 6 month mark:
http://www.e-cigarette-forum.com/forum/pdf/electronic-cigarettes-smoking-cessation.pdf

I don't think anyone would be surprised if they were found to be many times more effective at the long term 20 month mark - but that research has not yet been carried out, AFAIK.

 

[zoiDman]

Just curious. What do you think an acceptable margin of error would be for that 7% success rate? 1/2%? 1%? 5%? More?

 

[rolygate]

Not sure what you mean.

All statistics are perfectly accurate if it's your name at the top of the white paper. But if you're quoting it and you don't agree with the author, statistics are the work of the devil and as much as 1000% out. Depends on who's talking.



I'd love to know what the real number of smokers in the US is. It's variously quoted at 18% to 25% of the population. If the pop is 262 million, then 18% = 47m, 25% = 65m. David Goerlitz quoted around 60 million the other night, I think - which is about 23%. Mind you these all depend on the gen pop number being right, and as far as I can see some governments may underestimate that.

But I don't know if there is any reliable way to get the figure, which is basically a comment on stats in general - they depend entirely on how they were obtained and calculated.

On the other hand, website stats are of course absolutely perfect and accurate to within 0.1%.

 

[zoiDman]

Just for an very simplified example,

Say the "True" success rate for a given Population is Actually 4%. I do a study and claim that the success rate is 7% for this Population with a Margin of Error of +/- 5%. So my claim is Valid because the actual success rate falls between 2% and 12%.

Where as if I make a claim of 7% with a Margin of Error of +/- 1% then my claim is False because the True Population success rate is not within 6% to 8%.

The true art in Statistics is making an inference about a population with the least amount of error. The evil in Statistics is that results can be represented in just about any way you choose, given enough error.

There are many ways to reflect data in the way you want. A little more subtle technique is in choosing a sample and sample size.

 

[rolygate]

Ah, OK. Well, you'd have to go to the research papers directly and see what methods they used. I believe there are several that come in around 7%, but only one that says 2%, although there is a 50% chance I am wrong.

 

[zoiDman]

That’s all I trying to get across here. People need to read the “small print” in a study or a test to make an evaluation to just how reasonable the result are.

Here’s another one.

Say you want to show that a high percentage of people in the U.S.A. are for firearm ownership. So you poll 2,500 people at random and find that 89%, with a 1% margin of error, of the people believe they should have the right to own firearms.

A very Strong conclusion. A random sampling. A large sample size. A small margin of error. This must reflex the True Population. Right? The only thing I forgot to tell you is the 2,500 randomly surveyed were shoppers leaving a gun store in Texas.

So yes, reading how a survey or a test is conducted and to what error the claims are held to is a Very good thing.