Get Newsletter
AlzRisk Risk Factor Discussion
Back

Risk Factor:
  (anti-inflammatory medication, NSAIDs)
Risk Factor Type: Medications
Current Understanding:
The tables below summarize results from a modest number of reports that collectively show no association between use of non-steroidal anti-inflammatory drugs (NSAIDs) and the risk of Alzheimer disease dementia (AD) and total dementia. The results from observational studies are consistent with null findings from a randomized, double-blinded, placebo-controlled trial involving naproxen and celecoxib, which was halted early for safety concerns. While non-aspirin NSAID use (current, ever) assessed at the time of AD evaluation showed a protective association against AD risk, non-aspirin NSAID use assessed in years prior to AD evaluation showed no such association. Because by definition AD does not have an acute onset, but rather progresses over a number of years, the discrepancy in findings indicates that the time-updated analyses may not accurately reflect the association between NSAID use and AD risk. Studies that report time-updated results are particularly susceptible to bias from reverse causation, as individuals with cognitive decline may be less likely to be taking NSAIDs. Observational studies involving aspirin use show a marginally protective effect, although these results may be undermined by confounding by indication, healthy user bias, or selection bias. Because there are known risks to NSAID use, including cardiovascular events and gastrointestinal bleeding and ulceration, future studies examining the effect of NSAIDs by dose, duration, and timing on AD risk will likely be limited to observational studies. Because of these significant safety concerns and lack of evidence of any benefit, NSAID use is definitely not recommended to prevent AD. For a more in depth commentary, please see the Discussion.
Literature Extraction: Search strategy  * New *
Last Search Completed: 04 April 2015

Risk Factor Overview

Jackson JW, Kim IY, Arasaratnam MH, O’Brien J, Weuve J*, Blacker D*. “Non-steroidal anti-inflammatory drugs (NSAIDs).” The AlzRisk Database. Alzheimer Research Forum. Available at: http://www.alzrisk.org. Accessed [date of access].*

* contributed equally

* * *


Introduction

The tables on the Risk Factor Overview present a modest number of results from observational studies that, taken collectively, show no association between use of non-steroidal anti-inflammatory drugs (NSAID; including traditional broad-spectrum NSAIDs like ibuprofen and naproxen, cyclooxygenase-2 [COX-2] inhibitors like celecoxib and rofecoxib, salicylates [aspirin]) and risk of either Alzheimer disease (AD) or all-cause dementia. These results are consistent with the null results from a randomized, double-blinded, placebo-controlled trial of naproxen and celecoxib, but this trial does not provide definitive evidence because it was discontinued early for safety concerns and continued monitoring for AD for up to 5 years. Moreover, although non-aspirin NSAID use at the time of diagnosis was generally inversely associated with risk in observational studies, use of these agents assessed in the years prior to diagnosis was not. Because AD develops over many years and does not present acutely (as a stroke might), this contrast in findings indicates that the time-updated findings might not reflect the effects of NSAID use on AD risk. Results from observational studies of aspirin are more consistent with a marginally protective effect, but these results may be an artifact of confounding, healthy user bias, or selection bias, as discussed below. A note about terminology: most use of salicylates consists of aspirin use (i.e., acetylsalicylic acid; ASA), thus we refer to aspirin throughout this report except to specify a distinguishing feature or result pertaining to other salicylates or combinations of salicylate with other agents.


Potential Mechanisms of Action

Interest in a potential role of NSAIDs in AD prevention grew from a hypothesized role for inflammatory and vascular mechanisms in AD pathogenesis (see AlzRisk review of inflammatory biomarkers). The deposition of amyloid-beta plaques, tangles, and other insults can trigger inflammatory cascades of cytokines and other mediators, leading to neuronal cell death and further AD pathology (Fiebich et al. 1997; Ho et al. 1999; Pasinetti 2002). NSAIDs inhibit cyclooxygenase 1 and 2 enzymes (COX-1 and COX-2), which catalyze the synthesis of prostaglandins.

The two COX isoforms appear to play different roles in neuroinflammation. COX-1 (inhibited only by traditional NSAIDs such as ibuprofen and naproxen) promotes platelet aggregation and is expressed in microglia, which play a key initiating role in neuroinflammatory responses. COX-2 (inhibited by the selective COX-2 inhibitors like celecoxib and rofecoxib, as well as by traditional NSAIDs) is expressed in neurons and localized near synapses, within dendrites, and axon terminals (Wang et al., 2005), and appears to play a role in long-term potentiation and memory acquisition (Cowley et al., 2008). These data suggest that COX-2 may be involved in synaptic functioning and memory formation (Lyman et al., 2013). Another pathway could involve vascular and hemostatic effects through cerebrovascular pathology. Through inhibition of COX-1, aspirin reduces platelet aggregation—a key step in amyloid-beta deposition that releases inflammatory mediators (Casoli et al., 2013). Other lines of evidence include animal and cell culture studies, which suggest that NSAIDs may be associated with decreased burden of AD pathology (Lim et al., 2000).


Methodological Issues

Exposure

Exposure Classification. The studies included in this review varied in how they classified NSAID use. Some studies examined NSAID use overall, while others examined specific classes (e.g., COX-2 inhibitors) or particular agents (e.g., ibuprofen, aspirin). Studies also varied considerably in how they characterized the details of NSAID use. Many studies compared use vs. non-use, and some distinguished between current and prior use. These crude groupings omit details that may be critical for interpreting associations and, in many settings, may induce bias in the estimation of medication effects (see Confounding and Selection bias). Some studies assessed use only at baseline; others followed use over time, sometimes incorporating lags between use and time of diagnosis. A few studies with sufficiently detailed records or assessments examined categories of intensity and duration.

Reliability of exposure assessment. In most studies, information on NSAID use was based on self-report or reports from proxies. Several of these studies also examined pill containers of currently prescribed medications at interview visits. Other studies relied on medical records or comprehensive pharmacy dispensing records. For most agents, self-report methods of assessing medication use are the most liable to exposure misclassification (Strom et al., 2012), and errors in self-reporting appear to increase with older age (West et al., 1995; Wu et al., 2014). Moreover, this misclassification could vary by incipient disease status. In one scenario, persons aware of their cognitive decline might be more inclined to carefully report medication use; or perhaps, having more severely impaired memory, they may be more likely to erroneously report their use than those with normal memory function. Although the use of medical records and pharmacy data can sometimes counter errors from self-report of drug use, such records are problematic for NSAIDs because these databases do not capture over-the-counter (OTC) use of broad-spectrum NSAIDs and aspirin for acute or chronic pain and inflammation, or aspirin for cardiovascular prevention. A large prospective study of US adults without history of cardiovascular disease at baseline (ages 45-84 years) examined NSAID use through medical inventory and structured interview starting in 2000 (in the US, naproxen became available for OTC use in 1994); aspirin and ibuprofen use were mainly OTC (79%), whereas naproxen OTC use was much lower (about 32%) (Delaney et al., 2011). The National Health Interview Survey (NHIS) reported that 25% of adults ages 60-69 and 28% of adults ages 70 and older regularly used non-prescription aspirin in 2005, where regular use was defined as using aspirin at least three times a week over the three previous months. By 2010, the prevalence of regular aspirin users had increased to 41% for individuals age 60-69 and 46% for individuals 70 and older (Zhou et al., 2014). OTC use of non-ASA NSAID followed a similar increasing pattern. The high proportion of OTC use among NSAID users will usually result in exposure misclassification, particularly underreporting, for most studies, including those reviewed here.

Drug Class Effect. NSAIDs represent a group of drugs that provide analgesic, antipyretic, and in higher doses, anti-inflammatory effects. NSAIDs fall into three basic classes: traditional broad spectrum NSAIDs, COX-2 inhibitors, and aspirin. These are alternatively categorized as selective (COX-2 inhibitors) and non-selective agents (all others, but sometimes aspirin is distinguished from these). Because the pharmacological action of NSAIDs varies by type, many studies have analyzed aspirin and traditional NSAIDs separately. No clear differences in the associations between use of these agents and AD risk emerged in individual studies, but, in general, analyses of aspirin use yielded a marginally protective associations (the numbers of results for a given usage classification scheme [e.g., current use, ever use; baseline use, use at time of diagnosis] were insufficient for meta-analysis).

There is some speculation that a subset of NSAIDs within classes selectively lower amyloid beta peptide 42 (e.g., ibuprofen, piroxicam), and may be responsible for any observed AD risk reduction, but the epidemiologic results have been inconsistent (Szekely et al., 2008).

Duration and timing of exposure.. The duration of NSAID use may have important implications for the effect of therapy on AD risk. Few studies obtained sufficient information to study exposure duration in detail. Only a few of the studies reviewed here categorized exposure by duration of use, and some of these had findings consistent with a decrease in AD risk with longer-term use (in t' Veld et al., 2001; Stewart et al., 1997; Zandi et al., 2002). In addition, longer use typically implies earlier initiation, which also may be relevant. However, duration analyses can be extremely vulnerable to confounding by indication, because long-term users typically consist of individuals with conditions that share risk factors with dementia (e.g., persons with rheumatoid arthritis are long-term NSAID users and have higher risk for cardiovascular disease (Crowson et al.; Ong et al., 2013) (see Confounding)). Another limitation is that duration of use will partly reflect survival and discontinuation in response to gastrointestinal or cardiovascular events (see Selection bias).

Dosage. The therapeutic properties (and toxicity) of a given NSAID vary by dosage. Low doses of aspirin are sufficient for clinically meaningful antiplatelet activity, and low doses of aspirin and other NSAIDs confer analgesic and antipyretic effects. The anti-inflammatory property of all NSAIDs is not typically seen until relatively high doses are taken. No studies distinguished between dosage independent of duration, and like duration, dosage is tightly bound with indications that may themselves alter risk for AD. Studies that had information on the strength and amount of drug dispensed determined a standard daily dose and then classified users as heavy, intermediate, or light/non-users (Breitner et al., 2009; in t' Veld et al., 2001).

Confounding

Observational studies of NSAID use should be interpreted with caution in light of the potential for confounding by indication, contraindication, and characteristics of “healthy adherers.” Long-term users of NSAIDs may categorically differ from sporadic users and non-users, as both selective (e.g., Cox-2 inhibitors) and non-selective agents (e.g., aspirin and other COX-1 inhibitors) are regularly prescribed to treat inflammatory and pain-related symptoms of rheumatoid arthritis, which is associated with cardiovascular disease (Crowson et al., 2013; Ong et al., 2013). Furthermore, low-dose or "baby" aspirin is often used to prevent arterial thrombosis in persons with extant or high risk for cardiovascular disease, and persons who adhere to such regimens tend to be of higher socioeconomic status (SES) (Wallach-Kildemoes et al., 2013) and engage in other healthier behaviors that could reduce their dementia risk (Simpson et al., 2006; Yaffe et al., 2013). Prescription NSAID use may also be more common among persons with higher SES because of their better access to medical care. A substantial fraction of over-the-counter use, in contrast, involves low-to-moderate doses of aspirin or ibuprofen for relieving minor transient or chronic pain and inflammation. Along these lines, a study found that over-the-counter NSAID use occurred less often among those who were prescribed lipid-lowering medications, had a diagnosis of diabetes, were physically inactive, or had incomes below $25,000 (Delaney et al., 2011). The nature of the confounding by indication and by SES and cardiovascular disease, therefore, may differ by type of NSAID. In observational studies of NSAID use and AD risk, the extent of “channeling,” where certain kinds of persons only use certain types of NSAIDs for given amounts of time, is unclear. If, for example, long-term use occurs only among persons treated for arthritis or low-intensity chronic pain, or for cardio-prevention (i.e., low-dose aspirin), then analyses would have to discard persons without these conditions to better ensure a comparable group of non-users. In more severe scenarios, channeling could make analytical adjustment for confounding intractable.

Bias can also arise from the inclusion of prevalent users—persons who have used an NSAID before the study follow-up period began (Ray, 2003). Prevalent users represent a “survivor” cohort where use is implicitly “continued use” among persons who are still alive. Some traditional and selective NSAIDs increase risk for cardiovascular events (McGettigan and Henry, 2011), which are sometimes fatal and thus can shorten life-expectancy. Thus, including prevalent users in an analysis can lead to a survivor-type bias as described later (See Selection bias). The inclusion of prevalent users also complicates the control of confounding, because AD risk factors assessed at baseline may have been affected by prior NSAID therapy (particularly in the case of aspirin). Adjusting for these potential intermediates could lead to selection-bias, and failing to adjust for them would result in residual confounding. The studies reviewed here may be vulnerable to these sources of bias because none of the studies explicitly accounted for prescription or OTC NSAID use before study entry.

Reverse causation

Many of the studies followed their participants for many years. Some analyzed NSAID use at baseline in relation to AD risk over the subsequent follow-up, which averaged from a little over 1 year to 10 years across the studies included. Some studies that assessed NSAID use at each study cycle evaluated AD risk in relation to the most updated NSAID use status on file (i.e., time-updated analyses), typically NSAID use at the time of diagnostic assessment. With shorter gaps between NSAID use assessment and clinical disease, the influence of reverse causation—the effects of incipient disease on NSAID use—is plausible. Those becoming cognitively impaired may tend to express less pain, or be less likely to seek NSAIDs for pain, leading to overall less use and potentially an artificially protective effect of NSAIDs (Cole et al., 2006; Cole et al., 2011; Kruger et al., 2012). We would expect reverse causation to be particularly influential in analyses of NSAID use at the time of diagnosis, regardless of duration of use. Nearly all results from these types of analyses yielded inverse estimates of association with AD risk (RR<1.0), whereas results from analyses of baseline use yielded both inverse and positive (RR>1.0) estimates of association. Moreover, among the results comparing current or ever use with never or non-use, the interval between NSAID assessment and diagnostic evaluation was correlated with estimated association size (RR=0.38); i.e., shorter intervals corresponded to inverse associations. Nonetheless, two studies that examined use from zero to up to five years prior to symptom onset reported an inverse association with AD incidence irrespective of the lag size (in t' Veld et al., 2001; Stewart et al., 1997). Thus, it is unlikely that reverse causation completely explains the protective associations for aspirin or overall NSAID use.

Selection bias

It is possible that the studies reviewed here compared groups of persons who, because of different patterns of survival, discontinuation, or drug-switching, are fundamentally different in ways that could introduce bias. Regular NSAID use increases with age and occurs more often among persons with various medical conditions that are associated with mortality such as cardiovascular disease and cancer (Zhou et al., 2014). Because AD itself also shares risk factors for mortality (e.g. diabetes, see AlzRisk review of diabetes), a survivor bias could lower the estimated association between NSAID use and AD (i.e., overstate protection or underestimate harm). Adjusting for such risk factors would avert this source of bias, but many studies reviewed did not account for them in their design or multivariable-adjusted analyses.


Results from Other Lines of Research

Cognitive Decline

Several longitudinal studies have examined the association between use of NSAIDs and cognitive decline (Arvanitakis et al., 2008; Fourrier et al., 1996; Hanlon et al., 1997; Hayden et al., 2007; Jonker et al., 2003; Kang and Grodstein, 2003; Kang et al., 2007; Prince et al., 1998; Rozzini et al., 1996; Saag et al., 1995; Sturmer et al., 1996). Although two reported a benefit, most showed no association or even more rapid decline in cognition with use.

Clinical trials

Several randomized clinical trials have tested the efficacy of different NSAIDs for the treatment of AD (Jaturapatporn et al., 2012). Although an initial trial showed promising results (Rogers et al., 1993), subsequent trials did not show any benefit (Aisen et al., 2000; Scharf et al., 1999; Van Gool et al., 2001). The Women’s Health Study (WHS) found no effect of long-term use (mean, 9 years) of low-dose aspirin (100 mg, alternate days) on cognitive decline, despite a relatively long follow-up period (Kang et al., 2007). However, the aspirin dose used in the WHS was not strong enough to provide an anti-inflammatory effect. The Alzheimer’s Disease Anti-inflammatory Prevention Trial (ADAPT) trial, results of which are summarized in the table, was conducted to test the hypothesis that celecoxib or naproxen reduces the incidence of AD over seven years (Breitner et al., 2011; Lyketsos et al., 2007). The trial was terminated early (after two years average follow up) because of increased cardiovascular risks associated with celecoxib, as reported in other trials (in 2005, the US Food and Drug Administration [FDA] mandated that this drug carry a “black box” warning of these risks). The initial analysis suggested that neither celecoxib nor naproxen reduced AD risk and may have, in fact, increased it among persons in whom AD neuropathology was particularly advanced. Although continued observations of participants for an additional 18 to 24 months after randomization suggested a protective effect among asymptomatic individuals assigned to naproxen (Breitner et al., 2011), this finding was not borne out when the data were later examined after five years follow-up, yielding essentially null results (The ADAPT Research Group, 2013)(see results in theRisk Factor Overview). These latest findings agree with the observational evidence, which does not support the use of NSAIDs for AD prevention. The ADAPT trial results, however, are not definitive because the trial was designed to detect differences in AD incidence only after seven years, and during the first three years about 40% of persons discontinued their study medication; when the follow-up studies were conducted, only 12% of participants in each treatment group had been taking non-aspirin NSAIDs (not necessarily the same agents supplied in the trial) for six months or longer during the extended follow-up period. Such non-differential misclassification would tend to bias any estimated effect, harmful or protective, towards the null.

The types of individuals examined in the observational studies differed from those enrolled in the ADAPT trial, which has important implications for inference. Given the lack of medical-based exclusions in the observational studies’ analyses, these investigations centered on NSAID use in the general population, including “obligate” NSAID users such as those being treated for arthritis-related pain. The ADAPT trial, by contrast, required a family history of dementia for study inclusion but, for safety reasons, excluded persons who received anticoagulant therapy, used more than four doses per week of non-aspirin NSAIDs, or used more than 81 mg of aspirin per day (the “baby aspirin” often recommended for cardiovascular prevention in that time period). These criteria centered the inquiry on NSAIDs’ ability to prevent AD among high-risk but otherwise healthy individuals—who have no comorbid indication for regular NSAID use (ADAPT Research Group, 2009). Although the question of the NSAID’s effects on AD risk apart from other medical conditions is arguably the more relevant question, the removal of persons with conditions related to AD-risk resulted in lower than expected AD incidence during the initial study (0.7% observed versus an expected 2.5% per year) and reduction in statistical power (Breitner et al., 2011). Also, if the relation of NSAID use to AD risk is modified by family history of dementia, indications for long-term NSAID use, or associated characteristics, then the results from observational studies and ADAPT would not be directly comparable.

Systematic Reviews and Meta-Analyses

Other systematic reviews and meta-analyses of NSAID use and AD risk have appeared in the published literature (de Craen et al., 2005; Etminan et al., 2003; Szekely et al., 2004; Wang et al., 2015). The most recent of these (Wang et al., 2015) reported a significantly reduced risk of AD with NSAID use compared to no NSAID use. However, less stringent inclusion criteria were used, and little consideration was given to the timing of use relative to diagnosis (e.g., use at the time of diagnosis as opposed to use several years prior to diagnosis), so the summary estimates obtained partly reflect studies with prevalent cases, cross-sectional designs, and other features that we considered ineligible for ascertaining AD risk (see Inclusion criteria). Studies with these features are particularly susceptible to biases that may compromise validity (see Confounding, Reverse Causation, and Selection bias). These issues were not addressed in the review by Wang et. al. in their reporting of a protective association between both NSAID use or aspirin use and AD. An older review noted the strong potential for reverse causation, exposure misclassification, and publication bias (de Craen et al., 2005).

Other Analgesics

Some of the studies included in the tables also evaluated use of acetaminophen, a medication that, similar to NSAIDs, is used as an analgesic but that does not have anti-inflammatory properties (e.g., Stewart et al., 1997; Szekely et al., 2008a; Szekely et al., 2008a). Overall, acetaminophen was not consistently associated with risk of Alzheimer’s disease in these studies, and none of the findings was statistically significant. However, these analyses are susceptible to the same measurement, study design and analytical limitations as those of NSAID studies.


Discussion and Recommendations

On balance, the studies presented in the risk factor overview, including those from a large primary prevention trial—halted early for safety concerns—do not support the hypothesis that NSAIDs reduce AD risk. Whether any benefit exists for particular drugs, dosages, duration or timing of administration has not been fully explored, but several biases pose significant challenges for this line of research in observational data. In addition, there are known risks to NSAID use. In a large meta-analysis of observational studies, the largest increased risk for major cardiovascular events (summary RR>1.25) was observed for etoricoxib, etodolac, indomethacin, rofecoxib, celecoxib, diclofenac, and ibuprofen. Data were sufficient to evaluate the latter four by dosage. The CVD RR was even larger (summary RR>1.69), with high doses of rofecoxib, celecoxib, and diclofenac use, although low-dose use remained a predictor of risk. By contrast, ibuprofen use corresponded to increased CVD risk only at high doses (>1,200 mg) (naproxen was risk-neutral) (McGettigan and Henry, 2011). In addition, long-term use of traditional NSAIDs can cause gastrointestinal bleeding and ulceration (Allison et al., 1992), but the risk varies by agent and dose (Hernandez-Diaz and Rodriguez, 2000). The US Food and Drug Administration has issued black box label warnings to communicate the risk for cardiovascular events and gastrointestinal bleeding associated with prescription NSAIDS—both selective and traditional NSAIDs (US Food and Drug Administration, 2005). Overall, given the uncertainty in the evidence base regarding their benefit, and the well-documented safety concerns in elderly populations, long-term use for the prevention of AD cannot be recommended at this time. Because another randomized prevention trial is unlikely, future observational studies might improve the evidence base by incorporating more detailed exposure assessments and implementing more stringent exclusion criteria to reduce the potential for confounding by indication and selection bias.


References

ADAPT Research Group. Alzheimer's Disease Anti-inflammatory Prevention Trial: design, methods, and baseline results. Alzheimers Dement. 2009;5(2):93-104.

Aisen PS, Davis KL, Berg JD, Schafer K, Campbell K, Thomas RG, et al. A randomized controlled trial of prednisone in Alzheimer's disease. Alzheimer's Disease Cooperative Study. Neurology. 2000;54(3):588-93.

Allison MC, Howatson AG, Torrance CJ, Lee FD, Russell RI. Gastrointestinal damage associated with the use of nonsteroidal antiinflammatory drugs. N Engl J Med. 1992;327(11):749-54.

Arvanitakis Z, Grodstein F, Bienias JL, Schneider JA, Wilson RS, Kelly JF, et al. Relation of NSAIDs to incident AD, change in cognitive function, and AD pathology. Neurology. 2008;70(23):2219-25.

Breitner JC, Haneuse SJ, Walker R, Dublin S, Crane PK, Gray SL, et al. Risk of dementia and AD with prior exposure to NSAIDs in an elderly community-based cohort. Neurology. 2009;72(22):1899-905.

Breitner JC, Baker LD, Montine TJ, Meinert CL, Lyketsos CG, Ashe KH, et al. Extended results of the Alzheimer's disease anti-inflammatory prevention trial. Alzheimers Dement. 2011;7(4):402-11.

Casoli T, Balietti M, Giorgetti B, Solazzi M, Scarpino O, Fattoretti P. Platelets in Alzheimer's disease-associated cellular senescence and inflammation. Curr Pharm Des. 2013;19(9):1727-38.

Cole LJ, Farrell MJ, Duff EP, Barber JB, Egan GF, Gibson SJ. Pain sensitivity and fMRI pain-related brain activity in Alzheimer's disease. Brain. 2006;129(Pt 11):2957-65.

Cole LJ, Gavrilescu M, Johnston LA, Gibson SJ, Farrell MJ, Egan GF. The impact of Alzheimer's disease on the functional connectivity between brain regions underlying pain perception. Eur J Pain. 2011;15(6):568 e1-11.

Cote S, Carmichael PH, Verreault R, Lindsay J, Lefebvre J, Laurin D. Nonsteroidal anti-inflammatory drug use and the risk of cognitive impairment and Alzheimer's disease. Alzheimers Dement. 2012;8(3):219-26.

Cowley TR, Fahey B, O'Mara SM. COX-2, but not COX-1, activity is necessary for the induction of perforant path long-term potentiation and spatial learning in vivo. Eur J Neurosci. 2008;27(11):2999-3008.

Crowson CS, Liao KP, Davis JM, 3rd, Solomon DH, Matteson EL, Knutson KL, et al. Rheumatoid arthritis and cardiovascular disease. Am Heart J. 2013;166(4):622-8 e1.

de Craen AJ, Gussekloo J, Vrijsen B, Westendorp RG. Meta-analysis of nonsteroidal antiinflammatory drug use and risk of dementia. Am J Epidemiol. 2005;161(2):114-20.

Delaney JA, Biggs ML, Kronmal RA, Psaty BM. Demographic, medical, and behavioral characteristics associated with over the counter non-steroidal anti-inflammatory drug use in a population-based cohort: results from the Multi-Ethnic Study of Atherosclerosis. Pharmacoepidemiol Drug Saf. 2011;20(1):83-9.

Etminan M, Gill S, Samii A. Effect of non-steroidal anti-inflammatory drugs on risk of Alzheimer's disease: systematic review and meta-analysis of observational studies. BMJ. 2003;327(7407):128.

Fiebich BL, Hull M, Lieb K, Gyufko K, Berger M, Bauer J. Prostaglandin E2 induces interleukin-6 synthesis in human astrocytoma cells. J Neurochem. 1997;68(2):704-9.

Fourrier A, Letenneur L, Begaud B, Dartigues JF. Nonsteroidal antiinflammatory drug use and cognitive function in the elderly: inconclusive results from a population-based cohort study. J Clin Epidemiol. 1996;49(10):1201.

Hanlon JT, Schmader KE, Landerman LR, Horner RD, Fillenbaum GG, Pieper CF, et al. Relation of prescription nonsteroidal antiinflammatory drug use to cognitive function among community-dwelling elderly. Ann Epidemiol. 1997;7(2):87-94.

Hayden KM, Zandi PP, Khachaturian AS, Szekely CA, Fotuhi M, Norton MC, et al. Does NSAID use modify cognitive trajectories in the elderly? The Cache County study. Neurology. 2007;69(3):275-82.

Hernandez-Diaz S, Rodriguez LA. Association between nonsteroidal anti-inflammatory drugs and upper gastrointestinal tract bleeding/perforation: an overview of epidemiologic studies published in the 1990s. Arch Intern Med. 2000;160(14):2093-9.

Ho L, Pieroni C, Winger D, Purohit DP, Aisen PS, Pasinetti GM. Regional distribution of cyclooxygenase-2 in the hippocampal formation in Alzheimer's disease. J Neurosci Res. 1999;57(3):295-303.

in t' Veld BA, Ruitenberg A, Hofman A, Launer LJ, van Duijn CM, Stijnen T, et al. Nonsteroidal antiinflammatory drugs and the risk of Alzheimer's disease. N Engl J Med. 2001;345(21):1515-21.

Jaturapatporn D, Isaac MG, McCleery J, Tabet N. Aspirin, steroidal and non-steroidal anti-inflammatory drugs for the treatment of Alzheimer's disease. Cochrane Database Syst Rev. 2012;2:CD006378.

Jonker C, Comijs HC, Smit JH. Does aspirin or other NSAIDs reduce the risk of cognitive decline in elderly persons? Results from a population-based study. Neurobiol Aging. 2003;24(4):583-8.

Kang JH, Grodstein F. Regular use of nonsteroidal anti-inflammatory drugs and cognitive function in aging women. Neurology. 2003;60(10):1591-7.

Kang JH, Cook N, Manson J, Buring JE, Grodstein F. Low dose aspirin and cognitive function in the women's health study cognitive cohort. BMJ. 2007;334(7601):987.

Kruger TM, Abner EL, Mendiondo M, Schmitt FA, Smith CD, Jicha GA. Differential reports of pain and depression differentiate mild cognitive impairment from cognitively intact elderly participants. J Geriatr Psychiatry Neurol. 2012;25(2):107-12.

Lim GP, Yang F, Chu T, Chen P, Beech W, Teter B, et al. Ibuprofen suppresses plaque pathology and inflammation in a mouse model for Alzheimer's disease. J Neurosci. 2000;20(15):5709-14.

Lindsay J, Laurin D, Verreault R, Hebert R, Helliwell B, Hill GB, et al. Risk factors for Alzheimer's disease: a prospective analysis from the Canadian Study of Health and Aging. Am J Epidemiol. 2002;156(5):445-53.

Lyketsos CG, Breitner JC, Green RC, Martin BK, Meinert C, Piantadosi S, et al. Naproxen and celecoxib do not prevent AD in early results from a randomized controlled trial. Neurology. 2007;68(21):1800-8.

Lyman M, Lloyd DG, Ji X, Vizcaychipi MP, Ma D. Neuroinflammation: The role and consequences. Neurosci Res. 2013.

McGettigan P, Henry D. Cardiovascular risk with non-steroidal anti-inflammatory drugs: systematic review of population-based controlled observational studies. PLoS Med. 2011;8(9):e1001098.

Ong KL, Wu BJ, Cheung BM, Barter PJ, Rye KA. Arthritis: its prevalence, risk factors, and association with cardiovascular diseases in the United States, 1999 to 2008. Ann Epidemiol. 2013;23(2):80-6.

Pasinetti GM. Cyclooxygenase as a target for the antiamyloidogenic activities of nonsteroidal anti-inflammatory drugs in Alzheimer's disease. Neurosignals. 2002;11(5):293-7.

Prince M, Rabe-Hesketh S, Brennan P. Do antiarthritic drugs decrease the risk for cognitive decline? An analysis based on data from the MRC treatment trial of hypertension in older adults. Neurology. 1998;50(2):374-9.

Ray WA. Evaluating medication effects outside of clinical trials: new-user designs. Am J Epidemiol. 2003;158(9):915-20.

Rogers J, Kirby LC, Hempelman SR, Berry DL, McGeer PL, Kaszniak AW, et al. Clinical trial of indomethacin in Alzheimer's disease. Neurology. 1993;43(8):1609-11.

Rozzini R, Ferrucci L, Losonczy K, Havlik RJ, Guralnik JM. Protective effect of chronic NSAID use on cognitive decline in older persons. J Am Geriatr Soc. 1996;44(9):1025-9.

Saag KG, Rubenstein LM, Chrischilles EA, Wallace RB. Nonsteroidal antiinflammatory drugs and cognitive decline in the elderly. J Rheumatol. 1995;22(11):2142-7.

Scharf S, Mander A, Ugoni A, Vajda F, Christophidis N. A double-blind, placebo-controlled trial of diclofenac/misoprostol in Alzheimer's disease. Neurology. 1999;53(1):197-201.

Simpson SH, Eurich DT, Majumdar SR, Padwal RS, Tsuyuki RT, Varney J, et al. A meta-analysis of the association between adherence to drug therapy and mortality. BMJ. 2006;333(7557):15.

Stewart WF, Kawas C, Corrada M, Metter EJ. Risk of Alzheimer's disease and duration of NSAID use. Neurology. 1997;48(3):626-32.

Strom BL, Kimmel SE, Hennessy S. Pharmacoepidemiology. 5th ed. Chichester, West Sussex, UK: Wiley-Blackwell; 2012. p. p.

Sturmer T, Glynn RJ, Field TS, Taylor JO, Hennekens CH. Aspirin use and cognitive function in the elderly. Am J Epidemiol. 1996;143(7):683-91.

Szekely CA, Thorne JE, Zandi PP, Ek M, Messias E, Breitner JC, et al. Nonsteroidal anti-inflammatory drugs for the prevention of Alzheimer's disease: a systematic review. Neuroepidemiology. 2004;23(4):159-69.

Szekely CA, Breitner JC, Fitzpatrick AL, Rea TD, Psaty BM, Kuller LH, et al. NSAID use and dementia risk in the Cardiovascular Health Study: role of APOE and NSAID type. Neurology. 2008;70(1):17-24.

The ADAPT Research Group. Results of a follow-up study to the randomized Alzheimer's Disease Anti-inflammatory Prevention Trial (ADAPT). Alzheimers Dement. 2013;9(6):714-23.

Tyas SL, Manfreda J, Strain LA, Montgomery PR. Risk factors for Alzheimer's disease: a population-based, longitudinal study in Manitoba, Canada. Int J Epidemiol. 2001;30(3):590-7.

US Food and Drug Administration. COX-2 Selective (includes Bextra, Celebrex, and Vioxx) and Non-Selective Non-Steroidal Anti-Inflammatory Drugs (NSAIDs) 2005 [3/2/2014]. Available from: http://www.fda.gov/Drugs/DrugSafety/PostmarketDrugSafetyInformationforPatientsandProviders/ucm103420.htm.

Van Gool WA, Weinstein HC, Scheltens P, Walstra GJ. Effect of hydroxychloroquine on progression of dementia in early Alzheimer's disease: an 18-month randomised, double-blind, placebo-controlled study. Lancet. 2001;358(9280):455-60.

Wallach-Kildemoes H, Andersen M, Diderichsen F, Lange T. Adherence to preventive statin therapy according to socioeconomic position. Eur J Clin Pharmacol. 2013;69(8):1553-63.

Wang H, Hitron IM, Iadecola C, Pickel VM. Synaptic and vascular associations of neurons containing cyclooxygenase-2 and nitric oxide synthase in rat somatosensory cortex. Cereb Cortex. 2005;15(8):1250-60.

Wang J, Tan L, Wang HF, Tan CC, Meng XF, Wang C, et al. Anti-inflammatory drugs and risk of Alzheimer's disease: an updated systematic review and meta-analysis. J Alzheimers Dis. 2015;44(2):385-96.

Yaffe K, Falvey C, Harris TB, Newman A, Satterfield S, Koster A, et al. Effect of socioeconomic disparities on incidence of dementia among biracial older adults: prospective study. BMJ. 2013;347:f7051.

Zandi PP, Anthony JC, Hayden KM, Mehta K, Mayer L, Breitner JC. Reduced incidence of AD with NSAID but not H2 receptor antagonists: the Cache County Study. Neurology. 2002;59(6):880-6.

Zhou Y, Boudreau DM, Freedman AN. Trends in the use of aspirin and nonsteroidal anti-inflammatory drugs in the general U.S. population. Pharmacoepidemiol Drug Saf. 2014;23(1):43-50.