Trying to ballpark the market for cholesterol pills is becoming a pasttime. The latest survey of physicians shows that Lipitor will capture an incremental 10 percent to 20 percent of the statin market, which implies volume growth of 50 pecent to 70 percent, according to analysts at Leerink Swann, the Wall Street firm, which queried 61 specialists - 32 cardiologists and 29 primary care docs in the US.
The docs also forecast that 20 percent of Crestor patients will switch to generic atorvastatin, otherwise known as Lipitor, which translates to a drop of 3.4 percent of market share (down from 17 percent) over the next 12 to 24 months. Overall, up to 30 percent of the physicians questioned the need for branded statin therapy once generic Lipitor is broadly available, they write in an investor note.
For Crestor, the greatest reduction in utilization is expected to occur with the 10mg and 20mg dosage strengths. Vytorin usage is forecast to drop 10 percent. Following the introduction of Zocor generics, they continue, Lipitor sales dropped 29 percent in absolute terms over the three-year period following the launch of Zocor generics.
Based on historical data, Leerink Swann analysts Jason Gerberry and Seamus Fernandez write they expect roughly two-thirds of first year atorvastatin growth to occur in the second half of the generic launch year, which is when as many as eight generic alternatives will be available. They also point out that, "in the first year that generic Zocor became available, two-thirds of the growth occurred in the second half of the launch year, suggesting volume expansion is greater" when several generics are sold.
17 Comments
Nice to see additional information on this. I realized this was the case about a year a go looking at published prescription figures for the statins. Not only were the generics hurting the branded products, but the generics were responsible for driving an increase in the use (prescriptions) of statins overall.
With the introduction of generic atorvastatin we can expect to see not only losses for the remaining branded statins but also a sharp increase in the use of statins because the cost/benefit ratio is more attractive (acceptable?) to physicians, payers and patients. Whether this translates into better medical outcomes will of course be worth keeping an eye on.
J Boussart: Wellpoint recently reached an agreement with IBM for their gameshow computer Watson to "...to help medical professionals diagnose and sort out treatment options for complicated health issues."
I guess Wellpoint realizes that medical professionals might need some help, especially the ones who take statins and develop cognitive problems...sometimes referred to as the statin stupids...
It is amazing that few realize that there are no [zero] clinical trials ever done that show a mortality benefit from Lipitor, the same is true for Crestor [for heart deaths that is] and that there are no trials ever done showing any statin is a life extender for women [RR = 1.00 in 2 meta-analysis].
They imitate nitroglycerin [via the NO/eNOS pathway]. Fewer chest pain related non fatal events [in JUPITER minus about 50% but no decrease in cardiovascular deaths p=0.37].
All statins lower LDL if you think that is important, they all have side effects and in almost any study, they don't extend lives. Go for cheap if you must.
Ask your doctor how many like you would have to take x statin for how many years to have 1 death postponed .. For women and for anyone taking Lipitor, that number is infinity. For ALL statins, zero benefit before 20 months.
exceopt of course for the 2 year followup of ASCOT
http://eurheartj.oxfordjournals.org/content/early/2008/01/05/eurheartj.ehm583.abstract?maxtoshow=&HITS=10&hits=10&RESULTFORMAT=&fulltext=ascot-lla&searchid=1&FIRSTINDEX=0&resourcetype=HWCIT
John, that may be so [p=0.06] but at 1/2 study length and mean study end, the curves touch as to be drawn with a single pen. What happens AFTER a study is closed is anybody's guess and not evidence [especially when contradicted by placebo controlled phases in trials].
At ASCOT end there were TWO more "CV events" in women, not fewer, and the 49 in-house study [C Newman] ended with significantly increased mortality vs placebo [p=0.02 but low numbers and not considering the role of degrees of freedom].
SPARCL on placebo had fewer deaths at study end than on Lipitor. Pfizer reps are not allowed to say lives are extended. Pfizer knows this ain't so. They may suggest, imply or hint**), but they could get sued for stating so.
About ASCOT extension, and your reference: over 800 people died when they finally stopped looking. The p values for study end, extension and final recording for CARDIOVASCULAR deaths were 0.26, 0.30 and 0.13. Who are we kidding?
THE lowest P value at the 3 points was for "chronic stable angina" [proving the NO/eNOS effect of statins .. and resulting fewer revascularization presentations] are 0.007, o.12 and 0.002.
Importantly, the RR and P-value for heart failure are both 1.0 See your Table 4: total failure and the word NNT is not given in the EHJ article.
**) The ad: "Lipitor prevents heart attacks". The public thinks: "a heart attack can kill me" but fatal heart attacks are NOT ever prevented. No lies told but the truth abused. Life is our life's work is/was Pfizer's slogan ..
EV, maybe we are looking at different papers. The one I gave a reference to was p=0.02, not 0.06.
You've made a lot of arguments based on retrospective subgroup analysis. But if that's fair game, shouldn't Pfizer be allowed to claim a decrease in overall mortality in males, based on the highly statistically signficant effect seen in this subgroup in ASCOT? I don't think you would endorse Pfizer being allowed to make that claim. But the knife cuts in both directions, either retrospective subgroup analysis is valid or it is not.
Zocor reduced overall mortality in a highly signficant fashion in the 4S trial and in HPS.
The Lipitor CARDS trial just missed statistical significance on overall mortality (p=0.059).
The ASCOT trial was stopped early for ethical reasons when a 36% reduction in CV events was demonstrated, thus eliminating its chance to demonstrate an improvement in overall mortality within the regular trial protocol.
After that, the ethics of conducting placebo controlled trials became borderline at best, making it quite difficult to provide definitive evidence for a reduction in mortality for any drug but Zocor.
Lipitor was numerically equivalent to Zocor in the IDEAL trial in reducing overall mortality, but I'm not enough of a statistician to say whether the trial was large enough to prove non-inferiority, which was not a prespecified endpoint in any case.
Plenty of room for different opinions here, I just think your post implying that reduced risk of mortality for statin treated patients was a little one-sided. For most of these drugs the truth lies somewhere in between calling it a proven fact and calling it a sales rep's fanatasy.
John, we are looking at the same follow-up. To me, when 5000 are given Lipitor for mean study duration 3.3 years and the mortality curves TOUCH and there are 2 MORE CV events in women, that is failure. Moreover, ASCOT was a well done study up to 3.3 years.
Women are never a sub-group, they are included in studies and the majority of CVD deaths in the end, not men and for women, the zero mortality benefit from statin is medical fact as per 2 meta-analysis, JAMA and JACC [Medline 15138247 and 15337211].
Back to ASCOT follow-up, at mean 3.3 years they did the same trick EXCEL investigators did at 11 months when lovastatin's mortality increase reached a p value of 0.11: they discontinued the placebo group. In ASCOT's case, they "invited" physicians to offer 10 mg atorva to the placebo patients until end of the BP phase. Then, the p for mortality was 0.06. Then at some undefined later time, when the deaths had increased to 2.1x the 3.3 year number, the p was 0.02. So, let's say that was at mean 6.6 years and no placebo group for half of that, and count them, delta deaths were 62, 10/year, which makes a NNT(6.6) of 80, entirely ridiculous. Tellingly, endpoints in women remain unreported. As I said, CV mortality a wash with P's trending around 0.3.
You mention 4S [more deaths in women] and HPS: nothing significant in female deaths. You also mention "ethics". It is UNethical NOT to report full study endpoints including deaths [HPS] or to suggest doing longer trials would be unethical in a fatal disease when the statins are conclusive: no mortality benefit ever in women, in no studies before 1.5 years and in no studies at 5 years is there an increasing benefit vs placebo. Yet, most patients think a statin will prevent them from having 'the big one', that fatal heart attack. The urban myth.
The later statin trials include revascularizations, affected by NO/eNOS and thus mimicking nitroglycerin. Guess what, VIOXX vs placebo over 3.3 years would also lower 'revascularizations' of knee joints -- but the joints would be just as degraded. Ditto for statins and arteries: calcification continues at the same rate!
I think its a question of statistical power, EV.
The power of an outcomes trial is not determined by the number of subjects, but by the number of endpoint-defined events. For these small subgroups (to the extent you believe retrospective subgroup analysis tells you anything), you have to look not just at the nominal value but also the 95% probability interval.
If you look at the ASCOT paper for example, the nominal RR in women is about 1.1. But the 95% confidence interval for RR runs from 0.5 out to >1.5. This data does not show that lipitor decreases CV events in women, nor does it show that Lipitor does not decrease CV events in women. What it shows is that the trial was underpowered to answer the question.
So just to summarize a lot of long winded comments, I don't agree with your statement about "no effect in women" because 1) the trials were not powered to demonstrate that there was no effect, and 2) You seem to be using retrospective subgroup analysis in a selective fashion to argue that there is no effect in women, but don't accept that there is an effect in men, which is seen using an identical method of analysis.
I'm not sure that its fair to call terminating the trial at 3.3 years "a trick", this was done at the recommendation of the independent DSMB. This left Pfizer with a trial that was showing a trend toward reduced mortality that they would have liked to have for their label. I don't see how terminating the trial early was in Pfizer's interest.
I agree that the NNT for statins in primary prevention are quite high.
You've obviously spent a lot of time thinking about this, and its quite enjoyable to debate the issue with someone who has taken the time to become familiar with the data. Thank you.
EV, I'll just add that I don't have a lot of confidence in meta analyses. From my POV, Dr. Singh's CMAJ article on Chantix did a very elegant job of showing that the discretion exercised by authors in deciding which trials to include and how to do the statistical analysis allows them to drive the results to just about any conclusion they would like.
In the course of reviewing the data on this subject, I ran across several meta analyses of the effects of statins on mortality, all of which reviewed studies including over 150,000 patients, and presumably much of the same trials. Two concluded that there was a favorable effect on mortality, one concluded that there was none. I don't put much weight on any of them personally.
John, about the second paragraph of your 11:39 posting, I agree but that "under powered" argument is no longer valid:
1. If authors are going to use it on women, they should not include them in trials.
2. Since they did, 2 meta analysis taking ALL trials reporting mortality in women [trust me there were no others] they both conclude RR 1.00, i.e. NO mortality benefit. That included say 60,000 female patient-years on statin vs ditto placebo and: NADA. Let's say, 60 million for statin, 1/4 million doctors visits, 1 million pharmacy visits just to get the statin and this in the hope of living longer, without such effect. THAT is not underpowered, that is proof.
About Lipitor, when the reps are instructed not to state the stuff prevents deaths, that is telling -- and proof of what the trials tell us about men or women, high risk or low.
C Newman, a Pfizer employee, authored a n=~14,000 combo publication about 49 in-house studies but she forgot to mention mortality. It took the CEO of Pfizer to instruct his top man to release the numbers and yours truly to do the P value out of them: 0.02 for placebo doing better [taking 1 degree of freedom].
Full text Journal stuff I have published is here: www[DOT]health-heart.org/author.htm
P.S. I agree about meta-analysis in general but we're talking here about counting the dead bodies in placebo controlled trials. That is simple.
Your meta-analysis do not single out women and other risk groups and do not give numbers needed to treat per endpoint and there are many other things wrong with them [adjustments, conflicts, combining the uncombinable, etc.]
Thanks EV, this has been quite enjoyable and informative.
John, it would be nice if it were different: $130 billion for Lipitor and not a human living longer as per all trials.
The tragedy, it's destroying the health case systems and people's food budgets, and divert attention away from things and research that really works.
Q How many lipidologist can dance on an LDL globule? A That depends on the pharma money to make them dance.
Crestor's lower dose strengths are 5mg and 10mg. 20mg and 40mg are the higher doses.
"For Crestor, the greatest reduction in utilization is expected to occur with the lower dose strengths - 10mg and 20mg."
Well EV, I guess were just not going to agree on this one.
As for the pharma money/food issue, I'd say that if drug development provide the ROI attributed to it by folkson this board, pharmacompanies and venture capitalists would be increasing rather than decreasing spending on pharmaceutical R&D.
Legislation cutting drug prices by half would reduce NHE by 6%, a little more than one year's growth in NHE. The only way to have a real impact is going to involve taking a hard look at physician's and hospital fees, whic account for more than 60%.
Sorry John, there are 3 words in your post I don't get, ROI, NHE and folkson.
I just went through a brand new meta analysis full text. The abstract is here: www[DOT]ncbi.nlm.nih.gov/pubmed/21989464
Between the low and high power statin, the Relative Risk reduction for a "fatal myocardial infarction" is 0.96, more [increasing risk] for Lipitor and Crestor, less for the other statins. Neither are anywhere close to significance i.e. RR 0.96 but confidence interval 0.50-1.85. Total in those 8 trials 31 thousand hapless patients for 2 years.
Numerically the 2 "high power" statins do worse. I have my issue, like you, with any such analysis but where's the benefit; who are we kidding [and this includes in-house data].
Sorry EV, didn't mean to be opaque.
ROI is just return on investment. NHE is national healthcare expense.
Folkson is a typo, should have read folks on.
I'd prefer not to get into a discussion about meta analyses. I can't really attempt to rebut your points without digging into all the investigator's assumptions and choices in detail, and there are so many of them that it begins to resemble work more than a recreational debate. So I will neither concede the point, nor will I attempt to rebut any points you make regarding these.
John, I agree about meta-analysis and all of them have issues. The one I cited has a main author getting money from Pfizer an Merck -- not that there is anything wrong with that.
My beef if that no matter how you slice statins: they don't save lives. Pfizer's $b130 not withstanding. THAT is a problem.
ANY "health care" expenditure on statin is a waste and misplaced expenditure. Tragic but the for profit world we live in.