Ahrp.org

An Analysis of Use of Prozac, Paxil and Zoloft in USA 1988--2002
The IMR Patient Flow Model has been used to analyse the usage of the SSRIs Paxil, Prozacand Zoloft in the US since each drug was introduced. The results for Paxil were given in anopen letter to Congressmen Barton and Greenwood (Ref.1). The letter also introduced theIMR Model and summarised its development and capabilities. This report will answer threequestions with results produced by IMR.
1 How many US patients have consumed these drugs since date of introduction ? 2) How many US patients have continued to use these drugs for many years ? 3) How many US patients may have been killed by drug induced suicide in the first weeksof use ? Numbers of Patients.
The IMR model starts in the year of introduction and uses specific Inman usage profiles foreach drug (Ref. 2) The known quantity of medication has been consumed by approximately67.5M patients. Although it is possible that some patients may have switched betweendrugs and this would reduce the number of unique patients. Nevertheless a probable 8Mnew SSRI patients in 2002 indicates the magnitude of the potential harm, particularly asthese drugs have little or no proven efficacy above placebo for the majority of conditionsfor which they are now prescribed.
Table 1: New Patients on Prozac, Paxil and Zoloft.
New patients Starting on Drug in current year
1,547,442
1,547,442
1,947,338
1,947,338
1,787,521
2,001,310
1,020,216
2,373,784
4,244,205
2,863,022
1,467,859
1,677,421
6,008,302
2,280,438
1,289,065
2,171,122
5,740,625
2,328,614
1,653,982
1,940,201
5,922,796
2,622,047
2,106,261
1,675,933
6,404,241
2,663,209
2,037,645
2,106,666
6,807,520
1,855,358
1,861,142
2,068,905
5,785,404
1,380,479
2,330,549
2,292,718
6,003,747
1,610,645
1,887,306
2,463,645
5,961,595
1,805,015
3,046,058
3,041,536
7,892,609
27,037,820
18,530,071
22,025,721
67,593,612
An Analysis of Use of Prozac, Paxil and Zoloft in USA 1988--2002
Although the majority of patients are forced to drop out during the first year of use, manypatients remain on SSRIs for many years, some out of choice, but some become dependantand cannot withdraw. GSK have now been forced to admit that 25% (not 2% as theypreviously claimed) of patients will have “difficulty” in withdrawing (June 2003).
Other long term patients confuse their drug induced withdrawal symptoms and believe thatthese are their own original condition and that therefore the drugs must be making themwell. The lack of thoroughness of coroner investigations and the inadequacy of post-mortem analyses means that there is no data to indicate how many suicides may betriggered by attempted SSRI withdrawal. Table 2 presents the probable size of the longterm patient cohorts on each drug and their duration of dependence as at the end of 2002.
Table 2: Numbers of US patients in Long Term Use.
Duration of use by US Long Term Patients as at end 2002
Time on Drug
Number of Patients
1 year or more
2,831,468
2,565,975
2,894,978
8,292,421
3 Years or more
2,167,664
1,601,950
1,823,505
5,593,119
5 Years or more
1,669,161
1,142,992
3,757,579
7 Years or more
1,032,343
2,163,390
10 Years or more
12 Years or more
None of these drugs have ever been scientifically validated for such long term use either interms of efficacy or permanent central nervous system damage. The cost implications arephenomenal. For example, the cost of keeping 2.163 million patients on these drugs for 7years is over 20 billion dollars, (much of which are tax dollars). This money could havefunded alternative more effective and far less dangerous therapies. Table 2 is a snapshot atthe end of 2002, the whole picture is dynamic, the duration and cost of long term use willcontinue to grow as time passes, Paxil and Zoloft are relatively young so their long termpopulations and costs will increase.
This table illustrates the scope of IMR to analyse long term patient cohorts in detail for anySSRI. New patients, drug induced harm and costs can all be derived from one coherentbase, founded on the published work.
An Analysis of Use of Prozac, Paxil and Zoloft in USA 1988--2002
The IMR model uses some internal sequences to calculate the growing population ofpatients from the amount of medication that was consumed. But in order to calculate theprobable number of drug induced suicides it is necessary to have a suicide rate per givennumber of patients at risk. This suicide rate must be provided externally and must relate toactual trials and studies, it cannot be generated by IMR. In view of the serious implicationsof the tables 3 and 4 that follow, it is most important to briefly review the sources to giveprovenance to the chosen rates.
All new patients or volunteers on SSRIs regardless of their mental condition, child or adult,face a finite risk of drug induced suicide in excess of any risk of suicide that they hadbefore medication. GSK presented randomised clinical trial (RCT) results to the MCA forSeroxat in 1990 in their UK licence application. This showed that Seroxat was 8 times morelikely to cause suicide or suicide attempts than placebo giving a rate of 236 suicides per100K patients, zero for placebo. After some debate the application was withdrawn.
GSK then manipulated the results of the same trial, shifting suicides from Seroxat ontoplacebo, and re-applied in 1991. The adjusted figures showed that placebo was now twiceas dangerous as Seroxat, giving a rate for Seroxat of 168 suicides/100K patients and anincredible 361/100K for placebo. This affront to medical ethics, logic and common sensedid not dismay the MCA and on the basis of this data the MCA granted the lifetime UKlicence for Seroxat, which was never to be scientifically and critically reviewedsubsequently. This dangerous, inexplicable and incompetent approval in the UKundoubtedly influenced the FDA to follow suit 2 years later when the Paxil licence wasgranted.
In 1993, epidemiological studies (Ref.3) carried out by the Drug Safety Research Unit on acohort of 50 K patients who were using various SSRIs found gross rates of suicide asfollows:- 269 deaths/100Kfor Seroxat, 244/100K for Prozac and 173/100K for Zoloft. Theweighted average from the whole study is 219/100K. These gross figures must be reducedby an estimate of suicides that might have occurred in that population if they had neverreceived medication. This additional data, provided by a study on 460K people over 5years, suggests a cautious figure of 60/100K (Ref.4). The result is that the average SSRImay typically cause at least 160/100K suicides in excess of the suicides that may occurwithout medication.
Both clinical and epidemiological studies suggest that it would be quite justifiable to runIMR with an average SSRI induced suicide rate of about 160/100K. However, in theinterests of careful science and caution, a range of excess rates between only 32 to 104 druginduced suicides per100K patients has been chosen to predict the range of total druginduced deaths. To put it another way, the results have been calculated using rates of onefifth to three fifths of the excess induced suicide rates that have been measured in actualclinical trials and epidemiological studies on SSRIs.
Although all SSRI patients are at risk of induced suicide and other harm at every dosetransition (starting, stopping and increasing dose) the drug induced withdrawal suicides andmid treatment dose change suicides are not included in this analysis.
An Analysis of Use of Prozac, Paxil and Zoloft in USA 1988--2002
Table3 (Below): Drug Induced Suicides in the US ( rate of 32/100K patients ) Excess (Drug Induced) Suicides
32 deaths per 100K new patients
Table 4 (Below): Drug Induced Suicides at rate of 104/100K patients.
Excess (Drug Induced) Suicides
104 deaths per 100K new patients
An Analysis of Use of Prozac, Paxil and Zoloft in USA 1988--2002
Table1 shows that, in 2002, 7.8 million Americans became new users of either Paxil,Prozac or Zoloft and therefore they faced a net risk of induced suicide that is not likely tobe lower than 32/100K and very probably is higher than 104/100K. Nevertheless, evenusing these cautionary rates, somewhere between 2500 and 8200 excess suicides may haveoccurred in 2002 alone due to these drugs.
Whatever the value of the suicide rate, it was not zero. At least two thousand Americansmay have died in 2002 who were not warned of the possible lethal outcome of their SSRImedication. This is the consequence of a dysfunctional drug safety regulation system inwhich the drug manufacturers have not only failed to disclose evidence of great harm butalso have infiltrated virtually every organisation that could criticise, study and report ontheir drugs with scientific objectivity. This is compounded by the doctors who continuallyfail in their duty to report possible adverse effects, (i.e. suicide within days of starting thedrug), associated with the dangerous drugs that they prescribe so frequently and so readilyfor virtually all the problems of life, culture and society.
But the spotlight must now fall on the FDA (together with the other national regulators,) toexplain why it has been so unreactive, unaware and in such positive denial of the possibleharm and deaths caused to the American population by the drugs that it has approved since1988.
The significance of this paper is that for the first time some credible numbers are availableto assist the FDA who confidently assert, without any knowledge of the number of patientsat risk and the extent of harm that the “benefits of the drug outweigh the harm “. Now withthe IMR results “harm/ benefit ratios” can be discussed scientifically and three very seriousquestions can be asked.
1) What processes have the FDA used to measure the “benefit” for SSRI drug users?.
2) How many “units of benefit” equate to one drug induced suicide ?3) How many drug induced suicides will the FDA tolerate before robust regulatoryintervention is provoked ? Open Letter to Congressmen Barton and Greenwood “Number of US Citizens at risk to SSRIs” :Graham Aldred : 27 April 2004.
PEM Report No .6: Paroxetine : William Inman et al : 1993: Pharmacoepidemiology & Drug Safety: Vol 2 393-422.
A Comparison of paroxetine etc by observational cohort studies. FJ MacKay et al: 1997:Pharmacoepidemiology & Drug Safety: Vol 6 : Supp 3: 5-11.
Modelling Suicide Risk in Affective Disorders: A P Boardman & D Healy: 2001: Eur. Psychiatry vol16: 400-5.

Source: http://www.ahrp.org/risks/usSSRIuse0604.pdf