Alzheimer’s Disease Clinical Trial Decision-Making Among Patients with Mild Cognitive Impairment and Their Study Partners

Abstract

Background.

Prodromal Alzheimer’s disease (AD) clinical trials enroll patients with mild cognitive impairment (MCI) and their study partners. This study examined trial enrollment decision-making and risk for misunderstanding trial information presented in an informed consent process for prodromal AD trials.

Methods.

We performed structured interviews with patients with MCI and individuals who could serve as their study partners. We presented details of a hypothetical prodromal AD trial, followed by questionnaires to assess involvement of the study partner in trial decisions, patient capacity to consent, and patient health numeracy skills.

Results.

Among 65 patient participants, most were male (66%) and non-Hispanic White (88%) with a mean (standard deviation) age of 74.9 (8.4) years. Among 57 study partners, most were spouses (83%), female (74%), and non-Hispanic White (91%) with a mean (standard deviation) age of 70.3 (13.6). Most patient participants (66%) and study partners (72%) reported they would make enrollment decisions in partnership; 28% of patient participants said they would make the decision themselves, with input from a study partner. Twenty-five patient participants (38%) lacked capacity to consent to the trial; 38% demonstrated impaired health numeracy. Seventeen patient participants (26%) demonstrated both a lack of capacity and impaired health numeracy, among whom seven reported they would make enrollment decisions on their own, with input from a study partner.

Conclusions.

Some patients with MCI demonstrated a lack of capacity to consent, impaired health numeracy, or both, putting them at risk for misunderstanding trial information in the consent process for a prodromal AD clinical trial.

Background

Alzheimer’s disease (AD) clinical trials routinely enroll patients who meet criteria for mild cognitive impairment (MCI). MCI is defined as cognitive performance below that expected for a person’s age and education that elicits complaint but does not affect activities of daily living and does not meet criteria for dementia (Petersen and Morris 2005). Many trials exclusively enroll patients with MCI who demonstrate biomarker evidence of AD, also known as prodromal AD, due to their increased risk of progression to dementia and the opportunity to test disease-targeting treatments in participants exhibiting disease pathology. Participation in prodromal AD clinical trials involves co-enrollment with a study partner. Study partners are typically family members or close friends who can attend study visits with patient participants and provide information on their cognitive and functional abilities. The role of study partners is generally defined in the trial protocol, and there may be criteria for their participation based on the frequency or duration of typical contact with the patient participant.

In addition to the study partner requirement, prodromal AD clinical trials involve complex decisions, such as whether to undergo biomarker testing and learn the results, and whether to accept the risks associated with the investigational medication under study. For example, AD clinical trials of anti-amyloid therapies carry risk of amyloid-related imaging abnormalities (ARIA), and that risk is particularly increased for carriers of the APOE e4 allele (Hampel et al. 2023). Thus, deciding whether to enroll in a clinical trial that carries varying degrees of risk often requires higher-level executive functions.

Some patients with MCI may lack capacity to provide informed consent for prodromal AD clinical trials (Okonkwo et al. 2007; Jefferson et al. 2008). Previous research suggests patients with MCI frequently demonstrate impairment on functional tasks that require higher order cognition (e.g., financial abilities (Triebel et al. 2009; Griffith et al. 2003)) and may be prone to errors in risk calculation or misunderstanding of medication information (Zamarian et al. 2010; Pertl et al. 2014). Few data are available on how patients with MCI and their study partners approach the decision whether to enroll in AD clinical trials, which typically involve similar assessments of numeric information related to study risks. A pilot interview study with 23 dyads (Cox et al. 2019) found that patients with MCI may approach enrollment decisions differently based on the relationship of their study partner (i.e., spouse vs. another family member or friend). Whereas patients with a spouse consistently agreed that enrollment decisions would be made in close partnership, those without a spouse more frequently reported they would make decisions on their own.

Given the increasing focus on and complexity of AD clinical trials in early disease stages, further research is needed to understand how patients with MCI and their partners decide whether to enroll in trials. The objectives of the current study were (a) to elucidate how patients with MCI and their study partners make trial enrollment decisions, and (b) to identify patients at greatest risk for misunderstanding trial information in the informed consent process.

Methods

Overview

We performed structured research interviews with patients with MCI and their study partners from August 2019 to May 2021. Before the Covid-19 pandemic, interviews were performed in-person in patients’ homes or at the research center. During the pandemic, interviews were conducted remotely over the telephone or over a secure HIPAA-compliant video-conferencing platform. Interviews took approximately 30 to 60 min to complete, were audio-recorded, and were performed separately for patients and their study partners when possible to minimize potential influence on responses. All patients and study partners signed an informed consent form or provided verbal consent to participate in the study and received compensation for their participation. The study was reviewed and approved by the University of California, Irvine (UCI) Institutional Review Board.

Study population

Three UCI health system dementia specialists, including two neurologists (GTT and SAS) and one geriatrician (SPT), referred 176 patients who met Petersen criteria for MCI (Petersen and Morris 2005) during the 22-month period of the study. These criteria include (1) a cognitive complaint, (2) objective cognitive impairment for age and education, (3) preserved general cognitive function, (4) intact activities of daily living, and (5) not dementia. Newly diagnosed and returning patients with presumed underlying neurodegenerative etiology were eligible. Exclusion criteria were generally consistent with those for prodromal AD clinical trials and included other neurological disorders that can cause cognitive impairment, psychiatric diagnoses, and history of cancer (except for basal and squamous cell skin cancers) in the previous five years. Patients were contacted up to three times via telephone or email to provide information about the study and to schedule an interview. Interested individuals were asked to identify a study partner (i.e., a close family member or friend) who could also be contacted to participate in a separate interview, though availability of a study partner was not required for patient enrollment in the study. Hereafter, we use the terms “patient participant” and “study partner” to refer to patients with MCI and their family member or friend, respectively, who participated in the study.

Data collection

Data were collected and managed using Research Electronic Data Capture (REDCap) (Harris et al. 2009). Two researchers (CGC and CRS) performed structured interviews with patients and study partners from the same dyad. Before the interview, we administered the 30-item Montreal Cognitive Assessment (MoCA (Nasreddine et al. 2005)) or a modified 22-item telephone version of this test (T-MoCA (Katz et al. 2021)). We read scripted educational primers on the MCI diagnosis, AD biomarker testing, and clinical trials, followed by a description of a hypothetical prodromal AD drug trial that enrolls patients with MCI and their study partners (see Supplemental Digital Content, page 1). We then asked both patient participants and study partners how likely they would be to enroll in the trial (6-point Likert scale: extremely unlikely; very unlikely; somewhat unlikely; somewhat likely; very likely; extremely likely), and how they would approach the decision whether to enroll (5-categorical options: patient would decide without the input of their study partner; patient would decide with the input of their study partner; patient and study partner would decide in partnership; study partner would decide with input from the patient; study partner would decide without input from the patient) (Cox et al. 2019). Patient participants received additional questionnaires to assess their decisional capacity to consent to the hypothetical clinical trial and their health numeracy skills.

Decisional capacity

We used the University of California, San Diego Brief Assessment of Capacity to Consent (UBACC) (Jeste et al. 2007) to measure patient participants’ decisional capacity to consent to the hypothetical prodromal AD clinical trial. The UBACC has been validated in individuals with mild to moderate cognitive impairment who are considering AD clinical research participation (Seaman et al. 2015). The 10-item screening instrument includes questions focused on understanding and appreciation of the information in a clinical research protocol (see Supplemental Digital Content, page 3). Each item is scored on a scale of 0 to 2 points with 0 indicating a clearly incapable response, 2 indicating a clearly capable response, and 1 indicating a partially capable response. Total score range is 0 to 20 points. Prior to the start of the study, the research team prepared a list of answers for each item that would receive a score of 0, 1, or 2. After the scale was administered, two researchers (SAS and JDG) reviewed and agreed on final UBACC scores for each patient participant. Based on previous research, potential decisional incapacity was defined as a score of 14 points or lower (Seaman et al. 2015).

Health numeracy

To measure patient participant ability to assess numeric data, we administered the 12-item health numeracy assessment scale (see Supplemental Digital Content, page 3), which has demonstrated acceptable internal consistency (Cronbach alpha: 0.76) in previous studies of interpreting numerical health information in individuals with MCI (Delazer et al. 2013; Pertl et al. 2014). In patients with MCI, scores on this scale were shown to correlate with global cognitive performance, performance on mental arithmetic assessments, and measures of executive function that differ significantly from cognitively unimpaired age-matched controls (Pertl et al. 2014). The scale is scored by assigning 0 for incorrect answers and 1 for correct answers for a total possible score range of 0 to 12. Impaired health numeracy was defined as performance below a median score of 8, based on a previous study with age-matched controls (Pertl et al. 2014).

Sociodemographic characteristics

Patient participants and their study partners self-reported sociodemographic characteristics including age, sex (male or female), race (White, African American, Asian/Pacific Islander, Native American/Eskimo, Other), ethnicity (Hispanic, Non-Hispanic), years of education, marital status (single/never-married, married, separated, divorced, widowed), employment status (full-time employed, part-time employed, retired), and relationship of the study partner to the patient participant (spouse/partner, adult child, other, no one available). Race and ethnicity were collapsed into mutually exclusive categories (non-Hispanic White, non-Hispanic Asian, Hispanic, Other).

Statistical analyses

Descriptive statistics for patient participants and study partners were reported as mean (SD, standard deviation) for continuous variables and count (percent) for categorical variables. To understand how patient and study partner dyads would jointly approach the decision of whether to enroll in the clinical trial, we used the empirical joint distribution of complete dyad responses. In addition, we used the weighted kappa statistic to assess the overall level of intra-dyad agreement on the approach to the decision (i.e., patient would decide without the input of their study partner; patient would decide with the input of their study partner; patient and study partner would decide in partnership; study partner would decide with input from the patient; study partner would decide without input from the patient). We also assessed whether the reported approach to the decision was potentially sensitive to interview modality (in-person vs. remote) by looking at the marginal distributions of responses for patients and partners stratified by modality. We examined the frequency with which patient participants demonstrated lack of capacity to consent to the trial, impaired health numeracy, or both, based on the pre-defined cutoff scores for the UBACC and the health numeracy assessment.

Results

Of 176 patients referred to this study, we were unable to reach 61 and 50 declined participation. We completed interviews with 65 patient participants and 57 study partners (Table 1). Of the latter, 84% were a spouse or partner, 11% were adult children, and 5% had another relationship to the patient. Eight patient participants were unable or unwilling to identify a study partner who could participate in the interview, and one study partner participated without the patient for a total of 56 complete dyads. Forty-six percent of patient participants completed the interview in-person in their homes (n = 11) or at the research center (n = 19), and 54% of patient participants completed the interview remotely via telephone (n = 12) or video-conferencing (n = 23). The mean score for the standard in-person MoCA was 21.1 (3.4) out of 30, and the mean score for the remote T-MoCA was 16 (3.1) out of 22. Both are consistent with average scores among patients with MCI in previous studies (Nasreddine et al. 2005; Katz et al. 2021).

Table 1.

Characteristics of study participants.

Sociodemographic factors	Patients with MCI (N = 65)	Study Partners (N = 57)
Age in years, mean (SD)	74.9 (8.4)	70.3 (13.6)
Education in years, mean (SD)	16.8 (2.0)	16.6 (2.2)
Female sex, n (%)	22 (33.9)	42 (73.7)
Race and ethnicity, n (%)
White, non-Hispanic	57 (87.7)	52 (91.2)
Asian, non-Hispanic	4 (6.2)	4 (7.0)
Hispanic	3 (4.6)	1 (1.8)
Other race, non-Hispanic	1 (1.5)	0
Employment status, n (%)
Retired	54 (83.1)	46 (80.7)
Part-time work	7 (10.8)	5 (8.8)
Full-time work	4 (6.2)	6 (10.5)
Marital status, n (%)
Married/partnered	51 (78.5)	51 (89.5)
Divorced	8 (12.3)	0
Widowed	5 (7.7)	3 (5.5)
Single	1 (1.5)	3 (5.5)
Relationship to the patient, n (%)
Spouse/partner	-	47 (82.5)
Adult child	-	7 (12.3)
Other family/friend	-	3 (5.3)
Patient assessments
MoCA (n = 30), mean (SD) [range]	21.1 (3.4) [14–28]	-
T-MoCA (n = 35), mean (SD) [range]	16.0 (3.1) [8–21]	-
UBACC, mean (SD) [range]	15.6 (3.2) [7–20]	-
Health numeracy, mean (SD) [range]	8.5 (2.7) [0–12]	-

Abbreviations: SD: standard deviation; MCI: mild cognitive impairment; MoCA: Montreal Cognitive Assessment; T-MoCA: telephone-based MoCA; UBACC: University of California San Diego Brief Assessment of Capacity to Consent.

Two-thirds of patient participants were male, 88% were non-Hispanic White, 6% were non-Hispanic Asian, and 5% were Hispanic. On average (SD), patient participants were 74.9 (8.4) years of age and had 16.8 (2.0) years of education. Among study partners, 74% were female, 91% were non-Hispanic White, 7% were non-Hispanic Asian, and one study partner was Hispanic. Study partners were, on average, 70.3 (13.6) years of age and had 16.6 (2.2) years of education. Most patient participants (83%) and study partners (81%) were retired. We observed no meaningful differences in the distribution of sociodemographic characteristics between those interviewed in-person compared to remotely (data not shown).

When asked how likely they would be to enroll in the hypothetical prodromal AD clinical trial, half of patient participants reported being very likely (35%) or extremely likely (15%) to participate. Seventy-two percent of study partners reported being very likely (35%) or extremely likely (37%) to support the patient’s participation in the trial, and all but one said they personally would serve as the patient’s study partner in a clinical trial. When asked how they would approach the decision whether to enroll in the trial, most patient participants (66%) and study partners (72%) reported they would make the trial enrollment decision in partnership (Figure 1). Several patient participants (28%) and study partners (19%) said the patient would make the decision with input from their partner. Two patient participants (3%) and two study partners (4%) reported the patient would make the decision alone, all of whom were from non-spouse dyads. Two patient participants (3%) and three study partners (5%) reported the study partner would make the decision with input from the patient. No patient participants nor study partners said the study partner would make the decision alone.

Figure 1.

Patient (N = 65) and study partner (N = 57) report of who would make the decision whether to enroll in the hypothetical prodromal AD clinical trial.

When we stratified by interview modality, there were higher rates of patient participants and study partners reporting that the decision would be made in partnership when the interview was conducted remotely compared to those interviewed in-person (patients remote: 74% vs. in-person: 57%; study partners remote: 78% vs. in-person: 64%). Rates of reporting that the patient would make the decision with input from the study partner were lower for patient participants and study partners interviewed remotely compared to those interviewed in-person (patients remote: 20% vs. in-person: 37%; study partners remote: 16% vs. in-person: 24%). These differences are descriptive; we did not perform hypothesis testing on potential differences in the distributions of responses across interview modality due to the limited sample size.

Among the 56 complete dyads, 28 (50%) agreed that the decision whether to enroll would be made in partnership (Figure 2). Overall, 33 (59%) of the dyads agreed in their approach to the decision. In 15 (27%) dyads, the study partner reported they would have a larger role in the decision than the corresponding patient participant reported they would. In the remaining 8 (14%) dyads, the study partner reported they would have a smaller role in the decision than the corresponding patient participant reported they would. The weighted kappa statistic for dyadic agreement was 0.19.

Figure 2.

Empirical joint distribution of the dyadic decision whether to enroll in the prodromal AD clinical trial (N = 56 dyads).

Patient participants’ average (SD) score on the UBACC was 15.6 (3.1) with scores ranging from 7 to 20 points. Based on the 14-point cutoff, 38% of patient participants lacked capacity to consent to the prodromal AD clinical trial (Figure 3). Patient participants’ average (SD) score on the health numeracy assessment scale was 8.5 (2.7) with scores ranging from 0 to 12 points. Based on the 8-point cutoff, 38% of patient participants demonstrated impaired health numeracy skills. A quarter of patient participants (26%) demonstrated both lack of capacity to consent and impaired health numeracy, while half of the sample (49%) demonstrated neither (Figure 3). Among the 17 patient participants who demonstrated both lack of capacity and impaired health numeracy, seven indicated they would make the decision whether to enroll in the trial on their own, with input from a study partner.

Figure 3.

Proportion of patients with MCI (N = 65) who demonstrated lack of capacity to consent, impaired health numeracy skills, and both.

Discussion

Enrollment of patients early in the disease process in clinical trials has been critical to AD drug development (Sims et al. 2023; van Dyck et al. 2023). Relatively few studies, however, have investigated issues related to informed consent of patients with MCI, who make up a major portion of current trial activities (Parmigiani et al. 2022). The informed consent process is essential to ethical clinical trial conduct and relies on the ability of prospective patient participants to appreciate, understand, and apply reasoning to often complex medical decisions (Emanuel et al. 2000). Thus, this study sought to build upon previous research on decision-making in prodromal AD trials (Cox et al. 2019) to improve understanding of how patients with MCI and their study partners approach enrollment decisions, and to identify patients at risk for misunderstanding trial information presented in an informed consent process.

Consistent with previous research (Cox et al. 2019), we found that most patients with MCI and their study partners would make enrollment decisions in partnership. While relatively few patient participants and study partners in this study indicated the patient would make the decision alone, those who did were all members of non-spouse dyads. Dyad type impacts trial participation, with spouse dyads consistently overrepresented in AD clinical trial samples (Grill et al. 2013; Cary et al. 2015). Previous studies have found differences in availability, attitudes, and willingness to participate in research between spouse versus non-spouse AD care partners (Grill and Karlawish 2010; Grill et al. 2013; Cary et al. 2015). Moreover, patients from underrepresented racial and ethnic groups are more likely to have non-spouse study partners (Grill et al. 2013), suggesting the study partner requirement could contribute to underrepresentation of some groups in AD clinical trials. More research and specific strategies are needed to facilitate trial participation among patients with MCI who do not have a spouse, taking into account logistical, cultural, and structural barriers that may inhibit participation among diverse racial and ethnic groups (Karlawish et al. 2008; Watson et al. 2014; Largent et al. 2018). The current results suggest that such efforts, if they are successful, should be undertaken with simultaneous efforts to support patients who lack a spouse in the informed consent process, as they may more often insist on being self-reliant in considering trial participation. In fact, even in preclinical AD trials (in which participants are cognitively unimpaired), co-enrollment with a study partner has been recommended to ensure the safety and well-being of patient participants and the integrity of trials (Grill and Karlawish 2017).

More than a third of patient participants in this study demonstrated a lack of capacity to consent to the described trial. Similarly, over a third demonstrated impairment in health numeracy. A quarter of patient participants demonstrated both reduced capacity to provide consent and impaired health numeracy, potentially putting them at increased risk for misunderstanding information presented during the informed consent process and in the informed consent form. Previous studies have found that compared to adults without cognitive impairment, patients with MCI demonstrate higher frequency of compromised medical decision-making capacity, including appreciation, understanding, and reasoning, and that these consent standards become more compromised as patients progress to mild AD dementia (Okonkwo et al. 2007; Jefferson et al. 2008; Okonkwo et al. 2008). Impairment to higher order cognition in MCI can affect the complex decision to enroll in an AD clinical trial and subsequent decisions involved in trial participation, presenting challenges to the informed consent process. Specifically, several cognitive domains including short-term verbal memory and executive function (Okonkwo et al. 2008), perceptual speed (Han et al. 2015), and verbal knowledge (Stormoen et al. 2014), predict medical decision-making capacity in patients with MCI and may not always be observed clinically. Combined with these results, our observations serve as an important reaffirmation of the need to routinely assess capacity in prodromal AD trials, even in high functioning patients with MCI.

While many patients with MCI and their study partners express a desire to participate in prodromal AD trials, additional supports and assistance may be required to ensure patients at risk for errors achieve informed consent in an ethical manner. For example, Klein and Karlawish recommend assessing a patient’s capacity to consent, soliciting study partner consent together with patient consent, and using tools such as a one-page summary sheet to help patients with cognitive impairment process the trial information (Klein and Karlawish 2013). These supports may be especially important for patients with MCI who plan to make enrollment decisions on their own or plan to enroll with a study partner who may not have as close a relationship as a spouse. Given the length of prodromal AD clinical trials and progressive nature of the disease, additional assessments of capacity and informed consent checks at later timepoints in the trial may also be necessary. Future research is needed to examine the effectiveness of these different approaches to support the informed consent process in prodromal AD trials.

The framework of supported decision-making, as described by Largent et al. (2023), may be particularly relevant to promoting ethical conduct of prodromal AD clinical trials. Patients’ capacity to consent to trials is traditionally considered to span from adequate to inadequate. Surrogate decision-making – someone else making decisions for the patient – is a tool for patients who lack capacity. On the other end of the continuum, patients demonstrate adequate capacity and make their own independent decision. But in prodromal AD trials, especially complex trials that involve decisions based on contingent risk estimates (e.g., specific genotypes increase risk for adverse events), a middle ground may be more appropriate. Supported decision-making offers an ethical approach to the informed consent process by empowering patients with MCI to make their own enrollment decisions with the help of a trusted study partner. It also provides the option to tailor decisional support based on the patient’s impairment in specific cognitive domains. For example, study partners may help break complex enrollment decisions into smaller steps for patients who struggle with executive functioning. Additionally, trialists can develop decision aids to address different components of trial decisions, such as biomarker testing and risks of investigational treatments. Though research is needed to examine the efficacy of supported-decision making in prodromal AD clinical trial decisions, it may offer a promising approach to empower patient autonomy in these decisions.

Limitations

While this study adds to the limited literature on AD clinical trial decision-making among patients with MCI, there are important limitations. Patients and care partners were recruited from a single academic medical center and were a predominantly non-Hispanic White, college-educated sample, and thus not representative of the larger population of patients living with MCI and their family members. The small number of non-spouse dyads limited the opportunity to assess differences between the groups (i.e., spouse vs. non-spouse dyads). Compared to the consecutive clinical referral recruitment approach used here, future studies may wish to engage in stratified recruitment, to ensure key subgroups are adequately represented to assess for potentially important differences. Due to the Covid-19 pandemic, half of the data were collected via telephone or video-conferencing. Not all dyads interviewed remotely were able to be interviewed separately, which could have influenced patient participant and study partner responses to interview questions. We also observed some differences in results based on interview modality. Our data do not instruct whether this resulted from bias caused by the data collection approach or simple chance.

Conclusions

In this study, we found that most patients with MCI and their study partners were interested in AD clinical trial participation and would make the decision whether to enroll in partnership. Some patients with MCI were at risk for misunderstanding trial information, potentially requiring more assistance in the informed consent process. Research in larger, more diverse samples of patients with MCI is needed to understand how AD clinical trial decision-making varies across dyad types (i.e., spouse versus non-spouse), as well as in other groups underrepresented in trials, such as those from minority races and ethnicities (Gilmore-Bykovskyi et al. 2019). Such data will be critical to facilitate ethical and inclusive enrollment of patients with MCI and their study partners in prodromal AD clinical research to discover improved treatments for AD.

Supplementary Material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/23294515.2025.2576830.