Addressing the challenges to successful recruitment and retention in Alzheimer's disease clinical trials

Among the key challenges in Alzheimer's disease drug development is the timely completion of clinical trials. Unfortunately, clinical trials often suffer from slow or insufficient enrollment. Successful clinical trial recruitment describes a balance between expeditiously achieving full enrollment and ensuring an appropriate study sample. Investigators face a number of challenges to the successful negotiation of this balance. The failure to address these challenges means that drug development may take more time and money and that trial results may not adequately represent drug efficacy or may not be applicable beyond the study. We review the challenges to recruitment and retention in Alzheimer's disease clinical trials and present a framework to address them.


Trial design and conduct can aff ect recruitment
Success in meeting enrollment goals is not simply about advertising and outreach. Studies that are too long, require too many visits, or target enrollment of a population too diffi cult to recruit are in danger of slow or inadequate enrollment. In Table 1, we provide a literature summary of the rates of recruitment to a sample of multicenter AD trials. For these trials, we have calculated a summary recruitment rate statistic (RR) that is an approximation of the number of subjects recruited per study site per month for a given trial. Every trial faces unique challenges to recruitment, and every trial has its own recruitment goals. As such, comparisons among trials must be made carefully. Moreover, the data within Table 1 speak only to the rapidity with which a trial reached full enrollment. Timely fulfi llment of the proposed study enrollment is only one part of a truly 'successful' recruitment. Perhaps more important is the recruitment of a population of participants who are likely to complete the trial, are indeed affl icted with AD, and are representative of others with AD who will not be enrolled. Within a given trial, choices related to study design have a major impact on whether a trial achieves successful enrollment.

Visit frequency and study length
Decisions related to the total length of a study and the frequency of study visits are guided by study goals and often by concerns over safety. It is logical to expect that the longer the study and the greater the number of study visits, the greater the burden on participants and the more diffi cult recruitment will be. Trials of agents with high risk profi les or for which the risk profi le is largely unknown often require more visits to ensure patient safety. For example, early-phase studies (phase I or IIa) are often shorter (on the order of weeks to months) and require more frequent study visits than later-phase studies. Phase II AD trials of gamma secretase inhibitors have commonly used every-other-week study visits [1], making participation more daunting, espe cially for individuals who travel great distances to parti ci pate. In contrast, late-phase studies (phase IIb or III) that aim to evaluate effi cacy are commonly at least 18 months long. Th ese trials generally use study visits every 3 months.
Less commonly, the intervention itself necessi tates a more frequent rate of study visits. Ongoing trials of some immunotherapies for AD use medication infu sions once or twice per month.

Selection of the targeted Alzheimer's disease population
Th e target population is defi ned by the inclusion and exclusion criteria that participants must meet to enroll. Inclusion criteria should be designed to enroll only patients who truly suff er from AD and to maximize the likelihood of demonstrating a diff erence between drug and placebo when one exists [2]. Inclusion criteria generally identify a patient population of a specifi c disease severity. Th is is most often defi ned by a range of scores on the Mini-Mental State Examination (MMSE) [3]. Challenges in enrollment are not limited to trials of specifi c disease severities. As can be seen in the sample of published AD trials described in Table 1, examples of studies with fast rates of enrollment (for example, RR >1) exist for all disease severities. Similarly, slow enrollment can occur in trials in all stages of disease severity. Trials that fail to complete enrollment are also likely to go unpublished, given the probability that they will fail to meet the primary outcome [4]. To be clear, recruitment of participants with more severe disease faces unique challenges in comparison with studies of milder disease. Careful design and unique recruitment strategies, however, can be undertaken to overcome such challenges [5]. Besides disease severity, other specifi cations related to the population to be recruited can impact the rate of enrollment. For example, the Alzheimer's Disease Cooperative Study (ADCS) trial of estrogen replacement enrolled only women who had mild-to-moderate AD and who had undergone hysterectomy. Despite a somewhat wider range of MMSE inclusion criteria (12 to 28) than is typical, this trial enrolled only an average 10 subjects per month across 39 sites, and it took more than 3 years to complete enrollment [6].

Medication-related decisions
Criteria that exclude a large number of concomitant medications, though often necessary to ensure participant safety, can hinder recruitment. Many AD patients take supplements such as ginkgo biloba, and some trials exclude these patients. Alternatively, in trials examining available medications or supplements for therapeutic benefi t in AD, the greater availability of these agents can pose a challenge to enrollment. Th e TRIMCI study of the anti-infl ammatory agent trifusal in amnestic mild cognitive impairment (MCI) failed to meet its recruitment goals because of the high incidence of non-steroidal anti-infl ammatory drug use among potential participants, which was exclusionary [7]. A recent trial of latrepirdine (formerly dimebon) excluded patients taking medications currently approved by the US Food and Drug Administration (FDA) for the treatment of AD. Th is study was conducted in part in the US, where there is a high prevalence of use of these prescription medications among those diagnosed with AD. Th e data related to recruitment for this trial are not yet available.
To increase the appeal to participants seeking new treatments, some studies incorporate alternate allocation, whereby randomly assigned participants have a greater chance of being assigned to an active treatment group than the placebo group. Although this may increase the appeal of participation to some patients, alternate allocation also requires increased sample size to maintain statistical power and it remains unclear whether this strategy abbreviates the total study recruitment period [8].

Design changes made after study initiation
Changes to study conduct after trial initiation but before the close of enrollment can impact recruitment. Th e original entry criteria for a phase III trial of tarenfl urbil included mild-to-moderate AD patients with an MMSE score of between 15 and 26. Th ree months after enrollment began, the MMSE criteria for entry were changed to 20 to 26 as a result of fi ndings from a phase II study [9]. Overall trial recruitment occurred from February 2005 until April 2008. Such changes mid-enrollment can counteract previous recruitment strategies. Similarly, stopping a study medication dose prior to closing enrollment is likely to impact recruitment. Dosing changes, especially those brought about by safety concerns, must be communicated to new participants as part of informed consent and may deter enrollment of new subjects. Th e high dose of the anti-amyloid antibody bapineuzumab was halted for safety reasons prior to closure of enrollment in a recent phase III study. Alternatively, the publication of positive data related to a study drug might improve enrollment. Th e same phase III study of bapineuzumab was still enrolling when data were published from phase II effi cacy [10] and biomarker [11] trials. Data on the recruitment rates for the bapineuzumab phase III study are not yet available. Trials of drugs for which previous positive trials have been conducted are likely to enroll quickly. Th e initial trial of memantine in moderate-to-severe AD (MMSE 3 to 14) enrolled 252 subjects at 32 US sites over the course of 9 months [12] and demonstrated a signifi cant diff erence from placebo on the primary effi cacy outcomes. Consequently, a trial of memantine in moderate-to-severe AD (MMSE 5 to 14) patients taking donepezil effi ciently enrolled 404 patients at 37 trial sites over the course of 6 months [13].

Barriers to recruitment impact Alzheimer's disease patients and their study partners and shape trial populations
Successful trial enrollment faces many barriers, and most AD trials struggle to enroll. Th e ADCS trial of docosahexaenoic acid (DHA) enrolled 400 mild-to-moderate AD patients in 8 months, 10 months ahead of schedule, making it unique among AD trials. Th e agent tested in this trial funded by the National Institutes of Health was considered safe, allowing less restrictive inclusion and exclusion criteria. Th e trial also employed a 60/40 alternate allocation ratio toward active treatment. Th e factor that may have had the greatest impact on trial recruitment, however, was that it was conducted during a period in which few other trials in mild-to-moderate AD were recruiting and competition for subjects was minimal (Joseph Quinn, Oregon Health and Science University, Portland, OR, USA, personal communication).
As discussed, successful recruitment means more than just timely fulfi llment of enrollment goals. Trial participants should be representative of the greater AD popu lation. Th e mean age of participants in the DHA trial was 75.6 years. Fifty-three percent of participants were female. Th ese demographic factors are fairly represen tative of the greater AD-suff ering population. Partici pants in the DHA trial averaged 14.1 years of education. Th e over-representation of highly educated participants is common among AD trial populations [14] and stands in stark contrast to epidemiologic studies, which consis tently demonstrate that less than 12 years of education is a signifi cant risk factor for AD [15,16]. In the DHA trial, 90% of participants were Caucasian. Faison and colleagues [17] examined the race of AD trial participants, comparing 737 ADCS trial participants with 10,800 industry-sponsored trial participants. Th e authors found that only 10% of ADCS and 3% of industry-sponsored trial participants were non-Caucasian [17]. Given that African-Americans and Hispanics are at greater risk for AD than Caucasians [18,19] and that the proportion of AD suff erers who are of minority race or ethnicity will increase faster than that of Caucasians in coming decades [20], the low rates of minority enrollment in trials must be improved.
Among study partners in the DHA trial, 65% were female and 68% were spouses of the participant. Th e patient's primary caregiver most often fi lls the role of study partner and there are roughly 11 million persons in the US caring for a dementia patient. Th e majority of AD caregivers are women. Only a fraction of caregivers in the US, however, are spouses. Th e majority of caregivers are non-spousal family members, including primarily those who care for a parent or a parent-in-law [20]. Th e high representation of spousal caregivers in AD trials is striking and important. Trials off er patients and families an opportunity to feel active and involved in their medical care and in medical science's attempts to help them, others like them, and future generations. Many enroll in AD trials, however, in pursuit of therapeutic benefi t. Spousal caregivers may have greater motivation than do adult children caregivers to pursue new therapeutic options. Alternatively, there may be increased barriers to participation for adult children caregivers, who are more likely to be working full-time, more likely to have young families, and thus less likely to have the scheduling fl exibility to participate in clinical trials in the 9-to-5 clinic schedules in which they are generally conducted.
Th e overall diff erences between the enrolled population and the general AD population are troubling. Th ey suggest that the barriers to recruitment and retention signifi cantly shape the population under study and call into question the notion that the results of AD trials will be broadly applicable beyond a given study. We will next examine the various barriers to recruitment of AD trial participants, including the patient and the caregiver study partner.

Barriers related to the Alzheimer's disease patient-caregiver dyad
Th e decision to enroll in an AD trial is made by two people: the patient and their study partner. In this way, recruitment to AD trials is twice as diffi cult as recruitment to clinical trials that enroll only the patient. Th ose who choose to participate in a clinical trial commit signifi cant time and energy. Th is commit ment is justifi ed out of hope for personal and societal benefi t and trust in the investigator and study site [21]. Th e commitment is made with an understanding of given risks and requirements. Both the patient participant and the study partner participant must give informed consent and both must commit to full participation. Of course, patientcaregiver dyads cannot choose to participate unless they are aware of studies. At diagnosis, referral to trials is uncommon [22]. Th us, participation by those seeing physicians who do not personally conduct trials often requires active pursuit of information about study opportunities. Yet even when the patient and the study partner are aware of trials, they are still likely to encounter several barriers to trial participation. Th e barriers and facilitators of AD trial enrollment related to patients and caregivers are summarized in Table 2.

Barriers related to the Alzheimer's disease patient
Many AD patients who wish to participate in a clinical trial may not be eligible to do so. AD patients are, by defi nition, older. Older patients are likely to suff er from comorbidities that exclude participation. For example, current trials of immunotherapies exclude participants for a previous infarct (observed via magnetic resonance imaging [MRI]), even if it results in no neurologic sign or symptom. Given the high incidence of overlap between AD and vascular pathology [23], a substantial number of patients who might otherwise qualify for a trial may fail to be included, because of this criterion.
Older patients, especially in the US, also take a high number of prescription medications, which may similarly exclude participation. Even if these patients do not take an exclusionary medication at screening, trial protocols instruct investigators to enroll only patients whose medication profi les are stable and not likely to change through the course of the study. Protocols generally include patients taking AD medications, although these medications are subject to the same requirements. Th us, when enrolling mild patients who take only an acetylcholinesterase inhibitor (AChEI), the investigator is forced to consider whether to enroll the patient or start memantine (approved only in moderate-to-severe disease) to ensure stability through the course of the trial. Choosing the latter forces a delay in trial initiation and increases the likelihood that the patient will not be enrolled due to study closure or another reason.
Previous participation in an AD trial may exclude enrollment. Late-stage trials generally exclude participants of earlier-phase studies of the same drug. Similarly, most trials of active or passive immunization now exclude patients who have previously participated in a trial of any AD immunotherapy. In fact, for some ongoing trials, choosing to enroll means lifelong participation in one and only one trial. For example, the ADCS trial of nerve growth factor gene transfer aims to follow participants to autopsy and, given that subjects receive a therapy that is anticipated to deliver its therapeutic eff ect for as long as the neurons receiving it survive, being accepted into other trials is unlikely for recipients of the active therapy.
Some patients may be unwilling or unable to participate because of the procedures involved in a study. Individuals with pacemakers cannot undergo study-required MRIs and thus are excluded from trials that require imaging to ensure safety or use volumetric measures as mandatory outcomes. Many patients experience anxiety related to study procedures such as lumbar punctures. One phase II investigation of a gamma secretase inhibitor in pro dromal AD is enrolling participants in the randomly assigned treatment trial only if they meet specifi c criteria related to cerebrospinal fl uid protein analysis. Individuals unable or unwilling to undergo lumbar puncture are ineligible. Neuropsychological testing remains the hallmark of AD trials, co-primary outcome measures for all registration trials include one cognitive measure, and essentially all trials include a broad array of psychometric tests. For many trials, cognitive testing batteries are limited to the English language. Often, individuals not able to complete cognitive testing in the available languages at a study site are excluded. Such testing may require 3 to 5 hours to complete and can result in frus tra tion and distress for the participant [24]. In subjects aware of their impairment, the reminder of their cognitive struggles can be over whelming and may result in an unwillingness to participate.
Not all AD patients are aware of their impairment and those who lack insight may also lack the capacity to give informed consent. Some recent trials of aggressive therapies exclude individuals not able to demonstrate the capacity to provide consent. In these trials, the inability to comprehend trial-related proce dures and risks is a barrier to participation. Th e majority of AD trials, however, facilitate participation by permit ting a legally authorized representative to give the informed consent on behalf of a patient who lacks the capacity to do so for him-or herself. Most AD patients wish to be involved in the decision of whether to participate [25], and dyads that enroll in trials are likely to reach a joint decision. Although it is not clear how often it occurs, disagreement between patients and care givers about participation can be a barrier to enrollment [21].

Barriers related to the study partner
Patients who do not have a suitable study partner cannot be enrolled in AD trials. Th e study partner must be an individual familiar with the patient's medical and personal situation and the primary caregiver most often fi lls this critical role. At screening, the study partner provides an accurate medical history. Following enrollment, they provide transportation to study visits and serve as informants in a variety of study procedures and outcome measures. Between study visits, they monitor study and medication adherence. Th e role of the caregiver in the decision to participate in an AD trial is as important as that of the patient. Often, caregivers choose to participate in AD clinical trials out of hope for medical benefi t for the patient [21,[26][27][28]. Other motivations include despera tion resulting from a lack of other treatment options [21,26] and a desire to help medical science pursue a cure [21,26,27,29,30]. Trials off er the opportunity to interact with AD experts and access to new technologies that might not be covered by insurance.
If a study partner faces insurmountable barriers to participation, then it is unlikely that the patient will participate. Caregivers who decline participation cite a variety of factors that lead to their decision. Some caregivers cite the need to travel to the study site [21], and off ering car services to facilitate transportation or performing at least a portion of study visits in the home increases the likelihood that caregivers will support a decision to participate [31]. Individuals who report travel as a barrier, however, are not necessarily those who reside furthest from study sites. Th is suggests that, at least for some caregivers who decline participation, emotional and attitudinal factors about the logistics of travel play a large role.
Caregivers also face emotional burdens [32][33][34]. Th ey often cite the fear of side eff ects for the patient as a barrier to participation [21,28]. Many caregivers do not distinguish risks or benefi ts for the patient from risks or benefi ts for themselves [21]. Th e patient is most often a spouse or parent, and the caregiver does not wish to increase the patient's medical burden. Furthermore, increased medical burden for the patient is increased burden on the primary caregiver.
Finally, some caregivers cite the risk that the patient will not benefi t from participation as a barrier to enrollment [27]. Some caregivers who decline enroll ment cite doubts about the potential effi cacy of the agent under investigation as reason for refusing participation [28]. Th ese caregivers may defer partici pation in one trial to participate in another, more promising study. Th e same individuals are likely to cite the 'risk' of placebo as a deterrent to participation.

What factors impact trial retention?
Regulatory and ethical guidelines mandate that participants can withdraw their consent to participate in a clinical study at any time. Th erefore, good retention begins prior to enrollment, by recruiting study partici pants who are likely to complete a trial. Once trial conduct is initiated, making participation as convenient as possible for subjects and study partners optimizes retention.
Steps should be taken to inform participants of their value and the value of the research in which they are participating. Newsletters informing participants of trial progress can facilitate the feeling of being part of a larger agenda. For centers or investigators conducting multiple trials, annual luncheons honoring research participants can be eff ective retention tools, although these events must be conducted with sensitivity to participant confiden tiality and privacy.
A variety of factors can impact trial retention. Examples of trials that had poor retention exist, but often these trials faced extenuating challenges. A trial in mild-tomoderate AD of atorvastatin enrolled 98 participants, of whom 15 withdrew consent prior to random assignment 'primarily to participate in other trials' [35]. Similarly, the ADCS trial of dihydroepiandrosterone initially recruited 58 participants, but only 33 completed the 12-month trial [36]. Fifty-three percent of subjects randomly assigned to placebo dropped out of the study prior to completion, and the authors hypothesized that the high rate of dropout may have been the result of the widespread availability of FDA-approved AChEI therapies during study conduct [36].
We examined the retention rates in a sample of AD trials (Table 3). Although some trials may include a low dose without expectation of therapeutic benefi t, we chose to combine all active treatment doses in a single category.
In the very few occasions in which two active treatments were tested against placebo [37,38], we combined all active treatments in a single group. In cases in which participants completed a study off -medication, they have been included as completers whenever possible. Importantly, this summary is limited largely to trial reports in primary manuscripts rather than analysis of raw data and should be interpreted accordingly. Table 3 shows that the majority of subjects who enroll in AD trials are retained through trial completion and that, across disease severities, these rates do not substantively vary. MCI trials had an average retention rate of 71.6%, mild-to-moderate AD trials 77.7%, and moderate-to-severe and severe AD trials 75.4%. One might expect that, independent of disease severity, retention is easier in shorter trials. Even among some of the longest trials conducted, however, retention rates are high. Alternatively, some of the lower rates are for 6-month studies.
Few of the trials we sampled had a signifi cant diff erence between the treatment and placebo groups in the percentage of participants who completed the trial [13,39]. Th is supports the idea that altruism is a motivating factor for enrolling and continuing participation. If a patient or caregiver was interested in participation solely for the sake of gaining access to a new therapy, they might be likely to drop out of a trial if they concluded that they were randomly assigned to the placebo group (whether they were correct or not) or if they perceived that the patient is declining despite receiving study medication.
Recent analyses of the ADCS MCI trial of donepezil and vitamin E by Edland and colleagues [40] suggest that a variety of factors within a trial may indicate patients who will drop out prior to study completion. Th e authors found that the characteristics of participants who were likely to drop out were non-Caucasian race, less than high school education, and being unmarried (that is, having an adult child or child-in-law as a study partner). Furthermore, the analysis suggested that participants recruited to commercial trial sites (as opposed to academic sites) were at increased risk to drop out of a trial. Dropout rates at commercial sites were nearly double those of sites that were AD research centers funded by the National Institute on Aging [40]. In line with their analyses, in the trials that we reviewed, those with the largest study size (and as such were most likely to enlist non-academic sites) had the lowest retention

Outlook for the future
What is the available pool of patients for Alzheimer's disease trials?
Many promising therapies are in clinical development for AD and more will enter clinical trials in coming years. To plan the recruitment of adequate subject populations for these trials, it will be necessary to better understand the pool of eligible patients qualifi ed to participate. More than 400,000 Americans are diagnosed with AD annually [41]. Hence, investigators might assume that they have an ample (and growing) supply of participants for trials. Among all dementia patients, however, roughly half are moderately severe or more advanced in their disease [42,43] and therefore fail to meet the mild-to-moderate category for which most trials currently recruit. Th e majority of all AD patients are older than 75 years [43], increasing the likelihood of exclusion for reasons such as comorbidities or prohibited medications. In fact, analyses of general clinical AD populations suggest that only 10% to 13% are eligible for clinical trials [44,45]. In sum, the pool of eligible trial participants for AD trials is limited.  [1]). Th ere were more, though smaller, phase II studies. Screening ratios are generally better than 2:1 (2 patients screened to enroll 1) ( Table 1), but the fact remains that a signifi cant number of patients recruited will not be enrolled. Th us, if one considers the newly diagnosed patients each year, the barriers to enrollment, and the number of participants needed as multiple trials are conducted simultaneously, it is clear that the recruitment needs for AD clinical trials will remain a challenge that results in competition for eligible subjects. Strategies to overcome the current barriers to recruitment must be developed.

How can trial recruitment be optimized?
Th e most straightforward approach to improving the rate of enrollment is to increase the number of trial sites. AD trials have become increasingly 'global' , enrolling from multiple countries and continents within single studies. Th is change brings potential methodological [46] as well as ethical challenges when less industrialized nations are involved for which access to the drug (once it is approved) is not likely [47] (Declaration of Helsinki). Moreover, trial recruitment is diffi cult in all countries, not just the US [48]. It has been shown that, with increasing trial site number, the likelihood of placebo decline is reduced [49]. Furthermore, expanding the number of trial sites results in the inclusion of sites that are not focused on AD as a therapeutic area.
AD trials are optimally performed at trial sites experienced in their specifi c conduct by staff s well versed in the issues AD patients and their families face. Trials can be designed to facilitate participation for the subject and their study partner. Performing visits in the home and otherwise limiting travel hassles will increase the willingness to participate among caregivers.
Awareness of trials must be increased. Eff orts to increase awareness should target both patients and caregivers. Th e fact that most caregivers are adult children but most study partners are spouses indicates that there remains a large number of uninformed or unwilling potential participant dyads. Social media may provide an avenue to specifi cally target adult children caregivers. Th e Wisconsin Registry for Alzheimer's Prevention has successfully enrolled more than 1,400 middle-aged adult children of AD patients in a natural history study [50], and the use of television advertising has been an eff ective means of recruitment. Similarly, web-based patient registries such as www.patientslikeme.com [51] have been used eff ectively in other therapeutic areas. Internet use among those over the age of 65 is increasing, and 'wired seniors' are likely to seek health-care information specifi cally when online [52]. AD models of registries have been proposed and could target MCI and mild-stage AD potential participants and caregivers and also potential participants in prevention trials, such as baby boomers [53].
Th e advantage of disseminating the message of the value of clinical trials is to diversify the reasons why people enroll. Th e more reasons a person has for being in a trial, the more likely he or she is to enroll and, over time, stay in a trial. As participants experience adverse events and the eff orts of study visits, those participants whose reasons for enrollment include trust in the investigator and a desire to help others are more likely to stay in a trial than are patients whose sole motivation is benefi t to their health. Th is is especially true in AD clinical trials in which, to date, therapeutic benefi t has been elusive.
Eff ective methods to limit competition among trials, facilitate enrollment, and match eligible candidates with appropriate trials would improve recruitment and retention and result in a more rapid drug development. Agencies to fi ll such needs could have a major impact on clinical research. Such agencies, however, should be supported by federal and state governments, not forprofi t entities, and would thus be free of private corporate interests.

Conclusions
In summary, clinical trials in AD face a variety of challenges to recruitment and retention. Many trials struggle to complete enrollment in a timely fashion despite substantial eff ort. Trial enrollment may not repre sent the greater disease-suff ering population, and this may result in trial fi ndings that are not readily applicable beyond the study. AD trial retention is largely eff ective, although some variation does exist and is likely related to the population recruited and the caliber of study conduct. Responses to these challenges and improvement to recruitment and retention can and should be made. All eligible subjects must be made aware of participation opportunities, and trials must be designed in a manner that makes participation feasible for patient-caregiver dyads.