fb tracking

Letter on Ethical Concerns with the “Alzheimer’s Disease Anti-Inflammatory Prevention Trial” (ADAPT)

Tommy Thompson
Department of Health and Human Services
200 Independence Avenue, SW
Washington, DC 20201

Dear Secretary Thompson:

Public Citizen has uncovered major problems in a clinical trial, the “Alzheimer’s Disease Anti-Inflammatory Prevention Trial” (ADAPT), sponsored by the National Institute of Aging (NIA), part of the National Institutes of Health (NIH) and led by Dr. John Breitner at the University of Washington.  The problems we have found concern the scientific premise upon which the study is based (a premise brought sharply into question almost one year ago[1] and reinforced as recently as July 2002[2]), the selection of study drugs, the extraordinarily incomplete and misleading informed consent documents, and the promotion and recruiting practices. Because patients are taking drugs without true informed consent, Public Citizen urges the NIA to immediately stop this unethical trial and provide patients already enrolled in the trial with the information previously denied them on lack of plausible benefit as well as the possible health risks.

The NIA is funding a large (2,625 patient), long-term (seven-year) randomized study on the prevention of AD in people aged 70 or older who have a parent or sibling with serious age-related memory loss, senility, dementia, or AD. This trial is being conducted in six centers across the U.S.: Johns Hopkins University (Baltimore, MD), Boston University, University of Rochester, University of Washington (Seattle, WA), Sun Health Research Institute (Sun City, AZ), and the University of South Florida (Tampa, FL). Pharmacia and Bayer Pharmaceutical companies are providing the study drugs, two nonsteroidal anti-inflammatory drugs (NSAIDs): celecoxib (Celebrex; Pharmacia) and naproxen (Naprosyn; Bayer) and placebos. The study has been ongoing since January 2001, but new patients are still being recruited.[3]

We have done a search of the medical literature on AD and NSAID use and have discovered that the drugs being used in this study are not only unlikely to be effective but have the potential to inflict harm on these otherwise healthy individuals with little if any possibility of benefit. This information has not been provided to patients in the enrollment consent forms. It is possible that other NSAIDs might be beneficial, but not the ones chosen for ADAPT.

Three agencies under Health and Human Services (HHS) are involved in this trial: Center for Medicare and Medicaid Services (CMS, formerly HCFA), which is involved in recruiting, the Food and Drug Administration (FDA), which has responsibility for approving the protocol, and the NIA, which has provided funding and oversight. If the Medicare rolls and the imprimatur of HHS are to be used to recruit patients, it is of utmost importance that a.) the trials be on a reasonable scientific footing (not like here), b.) that full informed consent be obtained (not like here), and c.) that the recruitment letter makes it absolutely clear that enrollment is voluntary and that one will not lose Medicare coverage if one does not enroll (as done here).


AD is characterized, on autopsy of the brain, by extracellular amyloid protein plaques and intraneuronal tangles.[4]The origin of the extracellular material is a membrane protein, the amyloid precursor protein (APP), which is broken down to several smaller sizes, one of which, A-beta-42, forms aggregates outside the neurons in patients with AD. The amyloid hypothesis assumes that it is these plaques that lead to synaptic and neuronal injury.[5] Physicians use a variety of tests of mental functioning as a help in diagnosis, but there is no laboratory test; definitive proof of AD comes from autopsy.

The causes of AD are unknown. One hypothesis that gained support in the past was that AD was an inflammatory disease, since patients who, for other reasons, were taking anti-inflammatory drugs, especially NSAIDs, appeared to have a lowered risk of developing AD.[6] Thus, the idea evolved to test this hypothesis with a clinical trial giving NSAIDs. The NSAID hypothesis, however, has turned out to be more complicated than this, and it is now felt that the inflammatory response is secondary to amyloid deposition and not the cause of it.[7]

Newer data have provided some insight into how some NSAIDs might work to potentially protect against AD and to explain the puzzling lack of a class effect, i.e., not all NSAIDs appear to have a beneficial effect. This is because the effect of NSAIDs is “independent of their inhibition of cyclooxygenase,”[8] the basis of their normal anti-inflammatory action. Instead, the NSAID effect, if there is one, involves blocking the action of a different enzyme, gamma-secretase, one that plays a key role in forming the dangerous A-beta-42 itself. NSAIDs, by suppressing this enzyme activity, inhibit the breakdown of APP to form the plaques. Certain NSAIDs, but not all, suppress this enzyme activity such that less amyloid is formed. The fact that only certain NSAIDs inhibit this activity explains the lack of a class effect.[9] It would be logical, therefore, to select NSAIDs for AD prevention based on their gamma-secretase inhibitory effects, not on their cyclooxygenase inhibition. Unfortunately, this was not done. According to the investigators, the choice of celecoxib was “mostly because of pharmaceutical support”[10]; similarly, the choice of naproxen was also based on “Support from pharmaceutical company,”[11] while according to an NIA spokesperson, “we try to save money.”[12]

The information on how NSAIDs work came both from tissue culture and mouse models of AD. The NSAIDs shown to lower A-beta-42 production in these systems were ibuprofen, indomethacin, and sulindac but not celecoxib or naproxen, the two chosen for the ADAPT trial. Data from an observational human study, showing a protective effect of NSAID use[13], were recently reanalyzed to show that the apparent protective effect was restricted to ibuprofen, indomethacin, and sulindac, but, again, did not include naproxen.[14] A clinical trial of AD treatment, using rofecoxib (a different COX-2 inhibitor from celecoxib) and naproxen, was recently halted since neither NSAID showed any benefit in any of the endpoints.[15]

(There are two general types of NSAIDs: these work by inhibiting one or both cyclooxygenase enzymes [COX-1 and COX-2]. The older members of the NSAID class are not specific, inhibiting both enzymes [“COX-1 inhibitors”]; the newer members exhibit more specificity, inhibiting mainly cyclooxygenase-2 [COX-2 inhibitors].)


The ADAPT trial is being conducted at six centers in the U.S. It has three arms: celecoxib (Celebrex), naproxen (Naprosyn) and placebo. The primary objectives are to compare the efficacy of naproxen compared to placebo and celecoxib compared to placebo for the prevention of AD. Secondary objectives include determining if these treatments attenuate the cognitive decline sometimes associated with aging. 

Funding for ADAPT comes from the NIA.[16] The doses chosen are celecoxib, 200 mg twice a day and naproxen sodium, 220 mg twice a day. A summary of the study states that, “It is recognized that the doses of the trial’s two treatments are not pharmacologically equivalent (the celecoxib dose being somewhat stronger).”[17] (Italics added) One would think one would want equivalent pharmacological doses for a study of effectiveness (the celecoxib dose is the higher of the two recommended doses in the current label).


FAULTY HYPOTHESIS AS BASIS FOR STUDY: The inflammatory premise is one about which there has been much debate, and, although it received early support, appears to have been largely supplanted by more recent research.[18] In the current amyloid hypothesis, the inflammatory reaction is a secondary response, not the primary event.[19] NSAID effect is now thought to be due to an effect on gamma-secretase, not to an anti-inflammatory reaction. Although some NSAIDs inhibit the gamma-secretase, the two chosen for the ADAPT trial (naproxen and celecoxib) have never been shown to have such activity. They showed no lowering of amyloid levels in model systems,[20] no benefit in a population-based cohort study,[21] nor any benefit in a recent clinical trial of AD that used naproxen and rofecoxib, another COX-2 inhibitor.[22]


The use of long-term NSAIDs puts patients at risk for many adverse events. Yet, patients are not informed fully (or at all) about many of these risks: gastrointestinal (GI) adverse events, renal, hepatic, and hematological effects, fluid retention, cardiovascular events, preexisting asthma, geriatric use (increased likelihood of problems in the elderly), anaphylactoid reactions, and delayed bone and ligament healing. Patients, who are prescribed celecoxib and read the FDA-approved label (most patients do not have access to this), would see a section informing them of potential adverse events along with symptoms and instructions to report these to their physician (see Appendix 2 for part of this information). Below, we analyze the consent form for enrollment into the ADAPT trial to see how that compares to the FDA-approved product labels and recent scientific research.


This is a comparison of the Warnings and Precautions sections in the FDA-approved celecoxib[24] and/or naproxen[25] labels with the Risks/Discomforts section in the informed consent document. See Appendix 1 for text and side-by-side comparisons.

1) Gastrointestinal (GI) effects

Patients are told, “The study drugs can cause stomach irritation. This can range from mild stomach upset to ulcers” which “very rarely … bleed”.  The consent form never mentions perforation of the GI tract. The frequency of upper GI ulcers, gross bleeding, or perforation caused by NSAIDs, is stated to be 1%, yet the FDA-approved label says this is true only for those treated for 3-6 months (the label also states that the rate increases to 2-4% at one year with even higher incidences expected over longer durations of use). Furthermore, the celecoxib label states that only 20% of serious upper GI events are symptomatic, thus, the promise, “We will monitor you for any problems of this sort” does not protect patients from harm.

2) Renal effects

The FDA-approved celecoxib label warns that the elderly are at greater risk for kidney problems. ADAPT patients are not told that being elderly (as they all are) is an extra risk factor even if they have no problems with kidney, liver, or heart functions. There is recent data on COX-2 inhibitors that   “suggests that COX-2 inhibitors impair renal function and cause sodium retention in patients with mild pre-existing renal failure and presumably also in some elderly patients with, for example, volume depletion.”[26]

3) Hepatic effects

Although the FDA-approved labels warn of rare, severe hepatic reactions that can lead to death, ADAPT patients are only told that if they develop liver problems, they will stop taking the drug.

4) Hematological effects

Although the FDA-approved celecoxib label warns of anemia, the ADAPT consent form says nothing about this.

5) Fluid retention, edema, and blood pressure

Although the FDA-approved labels warn of the need for caution in patients with high blood pressure, fluid retention, or heart failure, ADAPT patients are only told that if they develop water retention, they will stop taking the drug.

6) Preexisting asthma

Although the FDA-approved labels warn of aspirin-sensitive asthma that can be fatal in susceptible individuals, ADAPT patients are told nothing about this potentially serious adverse effect.

7) Thrombotic tendency

The scientific literature provides a plausible biological mechanism that COX-2 inhibitors, like celecoxib, could increase the incidence of myocardial infarction.[27] COX-2 inhibitors do not share the beneficial antithrombotic platelet inhibition afforded by COX-1 inhibitors. Instead, COX-2 inhibitors block prostacyclin formation allowing thromboxane A2 production to proceed (thromboxane A2 can be a precursor of thrombotic events).[28],[29]

The placebo annualized myocardial infarction rate of 0.52% (derived from a meta-analysis of 23,407 patients) was significantly lower (p=0.02) than the 0.80% rate for celecoxib (CLASS trial).[30] ADAPT patients are told only that the doctors running the ADAPT trial do not think that there is a risk.

8) Delay in bone healing

The scientific literature has data showing that COX-2 inhibitors, like celecoxib, can prevent bone fracture healing in an animal model[31] with clearly important consequences. Since elderly people are at increased risk for fractures, they clearly need this information, yet ADAPT patients are told nothing.

9) Delay in ligament healing

The scientific literature indicates that COX-2 inhibitors, like celecoxib, can prevent ligament healing;[32] patients would need to stop medication in case of an injured ligament, yet ADAPT patients are told nothing of this risk.


AD presents a frightening prospect because it involves loss of mental functioning. Because the public has come to expect pharmaceutical solutions to disease, there is a great demand for preventive therapies for AD. Unfortunately, this trial is based on a hypothesis that is not supported by the latest research, thereby putting patients at risk of severe adverse reactions with little, if any, likelihood of a positive outcome. An especially vulnerable group is being exposed, i.e., the aged population (in this case, those 70 years old and older), who are apt to be less healthy than younger people and who will be exposed to a group of drugs (NSAIDs) for periods much longer than any period heretofore studied experimentally in any age group (most studies have ranged from three-months to one year in duration). Thus, there may be additional adverse events about which we are currently ignorant. More recently, the discoveries that COX-2 inhibitors block bone and ligament healing have added additional causes for concern, especially in the elderly.

The risks to which patients are exposed are either not clearly spelled out or are missing entirely in the patient consent forms. For example, the enrollment consent form mentions only stomach ulcers and “very rarely” bleeding ulcers (no mention of perforation) and gives the risk as 1% even though the FDA-approved label clearly states that the risk rises to 2-4% after one year of exposure and continues to increase as the duration of use increases.[33]

Even though a few NSAIDs may help in lowering A-beta-42 levels by inhibiting the enzyme which produces this AD-associated protein and, thus, might theoretically help to slow or prevent AD, the two NSAIDs being tested in ADAPT (naproxen and celecoxib) are not in that group. The choice of the two drugs for ADAPT was, unfortunately, heavily influenced by pharmaceutical support rather than basic science. A recent one-year AD trial testing naproxen and a different COX-2 inhibitor (rofecoxib) was stopped because neither drug showed any patient benefit on any of the endpoints.

Furthermore, many researchers in the field of AD research are concerned about the safety of long-term NSAID exposure and point out that “significant gastrointestinal and renal toxicity associated with long-term COX-1 inhibition limit the clinical utility of current NSAIDS as A-beta-42-lowering agents”[34](COX-2 inhibitors have not been shown to work in blocking A-beta-42 formation). Even the senior investigator for the ADAPT trial, Dr. Breitner, expressed concern over   “the potential to cause serious side effects when [NSAIDs were] used at high doses or over prolonged periods.”[35] Dr. Breitner continued, “Conventional NSAIDs can cause renal and gastrointestinal complications, including silent gastrointestinal bleeds that can be deadly, especially in the elderly”. As for COX-2 inhibitors, he says, “their safety when administered to older people for years has not yet been demonstrated”. Nevertheless, the investigators plan to expose elderly people to a COX-1 and a COX-2 inhibitor for 7 years, even though neither drug has been shown to function in either humans or model systems of AD.

Although COX-2 inhibitors were developed to have less GI toxicity, recent research does not support this.[36]   More importantly perhaps, they do not share the beneficial antithrombotic platelet inhibition afforded by COX-1 inhibitors but increase thromboxane levels that can lead to cardiovascular events.[37]

Even if there was a basis when the ADAPT trial was first proposed for believing that naproxen or celecoxib might have worked in preventing AD, there is no longer a scientific basis to support that hypothesis. We now know that only certain NSAIDs are candidates for a protective effect and that this effect does not depend on their inhibition of cyclooxygenase, as proposed in ADAPT, but functions rather through an inhibition of gamma-secretase.[38] Nevertheless, the Consent Statement for Enrollment states that, “Recent research suggests that they [celecoxib and naproxen] might [prevent AD] . . . “.[39]

The fact that there is no longer any biological basis for this trial, which is at the same time putting healthy elderly people at risk for a multitude of adverse reactions, provides the basis for our request that it be immediately terminated. This is reinforced by the failure to inform patients even of those risks mentioned in the FDA-approved labeling. The ADAPT trial should be stopped with the current 1000 patients (out of a planned total of 2625) and the patients enrolled should be fully informed about the extremely unlikely probability of efficacy of these two NSAIDs as well as the properties of these drugs that might put volunteers at risk of serious adverse reactions. There is no justification for continuation.


Elizabeth Barbehenn, Ph.D.
Research Analyst

Peter Lurie, M.D., M.P.H.
Deputy Director

Sidney M. Wolfe, M.D.
Public Citizen’s Health Research Group 



1)  GI Effects

FDA-Approved Labels


ADAPT Consent Form







“Serious gastrointestinal toxicity such as bleeding, ulceration, and perforation of the stomach, small intestine or large intestine, can occur at any time, with or without warning symptoms, in patients treated with nonsteroidal anti-inflammatory drugs (NSAIDS).” 


“ . . .upper GI ulcers, gross bleeding or perforation, caused by NSAIDs, appear to occur in approximately 1% of patients treated for 3-6 months, and in about 2-4% of patients treated for one year. These trends continue thus, increasing the likelihood of developing a serious GI event at some time during the course of therapy. (Italics added)


“Only one in five patients who develop a serious upper GI adverse event on NSAID therapy is symptomatic.” “To minimize the potential risk for an adverse GI event, the lowest effective dose should be used for the shortest possible duration.” [Bolding in label.]


“Serious gastrointestinal toxicity such as bleeding, ulceration, and perforation can occur at any time, with or without warning symptoms, in patients treated chronically with NSAIDs therapy.”


“Elderly or debilitated patients seem to tolerate ulceration or bleeding less well than other individuals and most spontaneous reports of fatal GI events are in this population.”

“The study drugs can cause stomach irritation. This can range from mild stomach upset to ulcers. Very rarely people with ulcers bleed from their stomachs.”


“Most people recover but a small number die from a bleeding ulcer. We do not think that taking celecoxib will increase your risk of a bleeding ulcer much, if at all. Taking naproxen may increase your risk to about 10 in 1000 [1%]. We will monitor you . . . If you develop any stomach problems, you will stop taking the study drug.”


2) Renal Effects

FDA-Approved Labels


ADAPT Consent Form






“Long-term administration of NSAIDs has resulted in renal papillary necrosis and other renal injury.“ “Patients at greatest risk of this reaction are those with impaired renal function, heart failure, liver dysfunction, those taking diuretics and ACE inhibitors, and the elderly.” “Clinical trials with CELEBREX have shown renal effects similar to those observed with comparator NSAIDs.”


“As with other nonsteroidal anti-inflammatory drugs, long-term administration of naproxen to animals has resulted in renal papillary necrosis and other abnormal renal pathology. In humans, there have been reports of acute interstitial nephritis, hematuria, proteinuria, and occasionally nephrotic syndrome associated with naproxen-containing products . . . “.

“In rare cases, the study drugs cause . . . kidney problems in people who already have some . . . kidney damage.” “If you develop . . . kidney problems, you will stop taking the study drugs. These problems almost always improve when the drugs are stopped.”

3) Hepatic Effects

FDA-Approved Labels

ADAPT Consent Form






“Borderline elevations of one or more liver associated enzymes may occur in up to 15% of patients.”


“Rare cases of severe hepatic reactions, including jaundice and fatal fulminant hepatitis, liver necrosis and hepatic failure (some with fatal outcome) have been reported with NSAIDs, including CELEBREX”.


“As with other nonsteroidal anti-inflammatory drugs, borderline elevations of one or more tests may occur in up to 15% of patients.”


“Severe hepatic reactions, including jaundice and cases of fatal hepatitis have been reported with naproxen . . . “.

“In rare cases, the study drugs cause liver problems in people who already have some . . . liver damage.” “If you develop . . . liver problems, you will stop taking the study drugs. These problems almost always improve when the drugs are stopped.”

4) Hematological Effects

FDA-Approved Labels


ADAPT Consent Form





“Anemia is sometimes seen in patients receiving CELEBREX.” “Patients on long-term treatment with CELEBREX should have their hemoglobin or hematocrit checked if they exhibit any signs or symptoms of anemia or blood loss.”


“Patients with initial hemoglobin values of 10 grams or less who are to receive long-term therapy should have hemoglobin values determined periodically.”

No information

5)  Fluid Retention, Edema, and Blood Pressure

FDA-Approved Labels


ADAPT Consent Form





“Fluid retention and edema have been observed in some patients taking CELEBREX.” “CELEBREX should be used with caution in patients with fluid retention, hypertension, or heart failure.”

“Peripheral edema has been observed in some patients receiving naproxen.”

“This [water retention] is a concern if you develop heart disease while you are in ADAPT. If this happens you will stop taking the study drugs.” 


“If your blood pressure goes up too much, we will tell you to stop taking the study drugs until it comes down.”


6) Preexisting Asthma

FDA-Approved Labels


ADAPT Consent Form





Celecoxib: “Patients with asthma may have aspirin-sensitive asthma. The use of aspirin in patients with aspirin-sensitive asthma has been associated with severe bronchospasm which can be fatal. Since cross reactivity, including bronchospasm, between aspirin and other nonsteroidal anti-inflammatory drugs has been reported in such aspirin-sensitive patients, CELEBREX not be administered to patients with this form of aspirin sensitivity and should be used with caution in patients with preexisting asthma.”


“Anaphylactoid reactions to naproxen, whether of the true allergic type or the pharmacologic idiosyncratic (eg, aspirin hypersensitivity syndrome) type, usually but not always occur in patients with a known history of such reactions.” “If such symptoms occur during therapy, treatment should be discontinued.”

No Information



Information for Patients (present in celecoxib label but not in informed consent):

“CELEBREX can cause discomfort and, rarely, more serious side effects, such as gastrointestinal bleeding, which may result in hospitalization and even fatal outcomes. . . patients should be alert for the signs and symptoms of ulcerations and bleeding and should ask for medical advice . . .”

“Patients should promptly report signs or symptoms of GI ulceration or bleeding, skin rash, unexplained weight gain, edema to their physicians.”

“Patients should be informed of the warning signs and symptoms of hepatotoxicity (e.g., nausea, fatigue, lethargy, pruritis, jaundice, right upper quadrant tenderness, ‘flu-like’ symptoms).”

“Patients should also be instructed to seek immediate emergency help in the case of an anaphylactoid reaction (see WARNINGS).”

“Patients with familial adenomatous polyposis (FAP) should be informed that CELEBREX has not been shown to reduce colorectal, duodenal or other FAP -related cancers, or the need for endoscopic surveillance, prophylactic or other FAP-related surgery.”.


[1] Weggen S, Eriksen JL, Das P et al. A subset of NSAIDs lowers amyloidogenic Abeta42 independently of cyclooxygenase activity. Nature 2001;414:212-216.

[2] International Alzheimer Congress, July 2002, www.alzforum.org/new/detail.asp?id=620

[3] Grant U01AG15477, Breitner JCS, Principal Investigator, Progress Report Summary, 3/1/01-2/28/02, p.163. http://www.alzheimers.org/trials/newtrials.html (accessed 8/26/02)

[4] Esler WP, Wolfe MS. A portrait of Alzheimer secretases-new features and familiar faces. Science 2001; 293:1449-54.

[5] Hardy J, Selkoe DJ. The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to therapeutics. Science 2002;297:353-356.

[6] In’t Veld BA, Ruitenberg A, Hofman A., et al. Nonsteroidal antiinflammatory drugs and the risk of Alzheimer’s disease. N Engl J Med 2001;345:1515-21.

[7] Hardy, op. cit.

[8] Ibid.

[9] Weggen, op. cit.

[10] Martin B. ADAPT Study Slides, June 26, 2001. Johns Hopkins University website accessed August 1, 2002.

[11] Ibid.

[12] Helmuth L. Protecting the brain while killing pain? Science 2002;297:1262-1263.

[13] In’t Veld, op. cit.

[14] International Alzheimer Congress, July 2002, www.alzforum.org/new/detail.asp?id=620

[15]International Alzheimer Congress, July 2002,  www.alzforum.org/new/detail.asp?id=623

[16] Dr. Neil Buckholtz, NIA, Project Manager for the ADAPT trial, says pharmaceutical companies only provided drug and placebo (telecom 5/31/02).

[17] ADAPT Protocol, Version 1.3, March 19, 2002, p.6.

[18] Selkoe DJ. Presenilin, Notch, and the genesis and treatment of Alzheimer’s disease. PNAS 2001;98:11039-11041.

[19] Hardy, op. cit.

[20] Weggen, op.cit.

[21]International Alzheimer Congress, July 2002, www.alzforum.org/new/detail.asp?id=620

[22]International Alzheimer Congress, July 2002,  www.alzforum.org/new/detail.asp?id=623.

[23] https://www.citizen.org/sites/default/files/enrollment.pdf

[24] Celebrex label, www.celebrex.com, October 2001.

[25] Naproxen label, Physician’s Desk Reference, 2001.

[26] Stubanus M, Riegger GAJ, Kammeri MC, et al. [Letter] Renal side-effects of cyclooxygenase-type-2 inhibitor use. The Lancet 2000;355:753.

[27] Mukherjee D, Nissen SE, Topol EJ. Risk of cardiovascular events associated with selective COX-2 inhibitors. JAMA 2001;286:954-959.

[28] Mukherjee D. Selective cyclooxygenase-2 (COX-2) inhibitors and potential risk of cardiovascular events. Biochemical Pharmacology 2002;63:817-821.

[29] Cheng Y, Austin SC, Rocca B, et al. Role of prostacyclin in the cardiovascular response to thromboxane A2. Science 2002;296:539-541.

[30] Mukherjee, 2001, op. cit.

[31] Simon AM, Manigrasso MB, and O’Connor JP. Cyclo-oxygenase 2 function is essential for bone fracture healing. Journal of Bone and Mineral Research 2002;17:963-76.

[32] Elder CL, Dahners LE, Weinhold PS. A cyclooxygenase-2 inhibitor impairs ligament healing in the rat. The American Journal of Sports Medicine 2001;29:801-805.

[33] Celebrex label, www.celebrex.com, October 2001.

[34] Weggen, op. cit.

[35] Zandi PP, Breitner JCS. Do NSAIDs prevent Alzheimer’s disease? And, if so why? The epidemiological evidence. Neurobiology of Aging 2001;22:811-817.

[36] Juni P, Rutjes AWS, Dieppe PA. Are selective COX 2 inhibitors superior to traditional non- steroidal anti-inflammatory drugs? British Medical Journal 2002;324:1287-1288.

[37] Mukherjee, 2002, op cit.

[38] Hardy, op. cit.

[39] https://www.citizen.org/sites/default/files/enrollment.pdf