The selection of participants for interventional microbiota trials involving cognitively impaired older adults

Abstract

Gut microbiota plays a significant role in nutrient extraction, metabolism, and immune function. Thus, the growing number of microbiome studies seek to link the presence and prevalence of specific bacteria, fungi, and viruses with a variety of physiological and disease outcomes. However, recruiting a diverse group of patients has been a challenge. Poor hearing and vision, lack of transportation, cognitive impairment, and a non-English primary language may interfere with patient enrollment as well as adherence to the requirements of a Microbiome study. Much of what we do know about diseases in older adults comes from studies that exclude many of these patients commonly encountered in clinical practice. The purpose of this review article is to highlight recruitment and retention strategies for engaging people who typically do not participate in microbiome studies, and it seeks to develop and explicate inclusion and exclusion criteria to promote more robust study results.

Introduction

A wealth of studies supports the role of gut microbiota on human functioning, including nutrient extraction, metabolism, and immune health [1]. Microbiome studies seek to link the presence and prevalence of specific bacteria, fungi, and viruses with a variety of physiological disease outcomes. This is particularly important in examining ailments with little or no curative or pharmacological treatments, such as Alzheimer’s disease and related dementias (ADRD). Multiple studies are beginning to highlight the association between specific microbiota and ADRD [2, 3]. Microbiome in Aging Gut and Brain (MiaGB) Consortium is a Florida state-wide consortium focused on developing novel data to promote understanding of how microbiome changes are linked with aging and ADRD [4–7]. The ultimate goal of these gut microbiome studies is to identify potential new variables that can be developed into new targeted interventions for the prevention or retardation of the ADRD process.

Degenerative neurological diseases, such as ADRD, have a long mean survival time of 7.6 years from the onset of clinical symptoms and likely have a preceding pre-clinical period of one or more decades [8]. Recent advances in the validation of amyloid plaque positron emission tomography (PET) imaging and in the development of serologic-based diagnostics, such as plasma phosphorylated tau, offer new opportunities for the identification of research participants at the early or even pre-clinical stage of ADRD [9].

Unfortunately, clinical studies often exclude many older adults commonly encountered in clinical practice, such as those patients who are cognitively impaired or are homebound. The homebound older adult population is estimated to be three times larger than the equally impaired and chronically ill nursing home population [10]. Based on data from 2011, the homebound population is also older, more likely to be non-White, and less educated than non-homebound individuals [11]. While research suggested that racial minorities such as Black and Hispanic people have an increased prevalence of ADRD, they are less likely to be represented in clinical research trials [12, 13]. As such, there is a selection bias against the inclusion of homebound and minority older adults in ADRD interventional studies [14]. Furthermore, the underrepresentation of minority groups in microbiome studies may reflect research participants’ fears of exploitation in biomedical studies and distrust of researchers [15, 16].

In addition, older adults with ADRD face many challenges. Poor hearing and vision capabilities, lack of transportation, socioeconomic challenges, and a non-English primary language may interfere with patient enrollment [17]. Microbiota studies represent additional challenges for these individuals. Compliance with stool collection, storage, and shipping instructions may be more complicated for older adults, especially those with cognitive and sensory deficits, and functional impairments. Detailed information on dietary, medication, and medical history is needed to determine potential biological and social determinant variables that may impact the gut microbiome.

Developing recruitment and retention strategies

Inclusive participation in microbiome studies must be a priority for researchers to mitigate disparity in research and subsequent health inequities. Developing inclusive prospective trials for preventative ADRD interventions is challenging. For studies that involve cognitively impaired older adults, recruitment initiatives may include home visits, local community center outreach, and church-based efforts. Networking with local research staff and community health workers can provide a bridge for researchers into the community. A partnership with a trusted advisor to the community of interest decreases fears about exploitation. Additionally, this person can meet potential participants in their local community, be flexible with appointments, and allow sufficient time for potential participants to ask questions and share concerns in their preferred language [18].

Focusing recruitment efforts on adult day care centers and assisted living facilities that serve minority communities would provide efficiencies in the recruitment of both underserved participants with ADRD and non-cognitively impaired older adult controls. The reliance on adult day care and assisted living facilities for enrollment needs to account for common diets and meal preparation among the participants at each facility. Whether this factor would be a source of bias or even a possible control factor remains unclear. Partnering with local medical neurologists can be a successful method to recruit participants. While many medical providers may be too busy to recruit, they can identify and invite potential participants, and their office staff can follow-up with potential participants. Collaborating with Medicaid-managed care, long-term care companies may provide access to large numbers of underserved potential participants. Other organizations, such as the Alzheimer’s Association and the Department of Elder Affairs, are trusted resources that can provide assistance in recruiting diverse populations. Another potential model for the recruitment of homebound older adults is the Department of Veterans Affairs home-based primary care programs (HBPC). HBPC provides comprehensive, interdisciplinary, longitudinal primary care in the homes of Veterans with complex medical, social, and behavioral conditions for whom routine clinic-based care is not effective [19]. Thirty-nine percent of HBPC enrollees have a diagnosis of ADRD [19]. The comprehensive electronic health record of HBPC enrollees contains detailed information regarding the use of medications and nutritional supplements, prior medical/surgical history, and quarterly dietary and mental health evaluations. As such, the electronic health record can also be leveraged for the prescreening of a large number of geographically isolated participants. Recruitment metrics for underserved populations should include potential participant interests, screening to enrollment ratio, and source of recruitment (recruiting personnel, physician, older adult facility, and clinic/hospital).

In longitudinal studies, retention strategies are equally important, as the loss of participants over time in ADRD clinical trials is not uncommon. For example, recent clinical trials involving IgG1 anti-AB monoclonal antibody therapies have a study dropout rate between 20 and 25% [20–22]. Older adults in low-income minority communities may frequently change phone numbers and addresses and may or may not have email. Regular communication with research staff, community health workers, and participants is critically important. Implementing multiple options for community-based touchpoints will mitigate study dropout due to loss of contact. Feedback to the community about general study findings may prevent discouragement while mitigating fear of exploitation in biomedical studies [23]. To promote retention, it is important to reduce participant burden through increased flexibility in appointments and meeting locations, using simple language in participant-required record keeping, and providing incentives for participation [24]. Additionally, the retention strategy should be culturally and socially appropriate and include considerations such as demographics, personal health beliefs, and cognitive status [24, 25]. Including access to additional resources such as health coaching, social support, and coping strategies for participants and their caregivers can be a valuable way to build relationships and trust with the participants and mitigate exploitation fears.

Retention metrics for underserved populations should include participant demographics (age, gender, race, ethnicity, education, etc.), personal health beliefs, and cognitive status [24]. Additionally, retention metrics should include drop-out rate (how many participants withdrew from different demographic and ethnic groups) and follow-up completion rate (how many participants actually returned for a follow-up visit each year). It is also important to include reasons for attrition and withdrawal, such as whether they decided not to participate anymore due to long visits, lack of transportation, medical issues, or something else. Was it something that the research team could have done better? Were the reasons different for individuals from different groups? It is important to regularly ask participants about their experience in the program and seek ways to actively improve their experience throughout their study.

Ethical considerations

The inclusion of underserved individuals with ADRD in microbiome studies is essential. Excluding them results in a lack of evidence-based research that fails to mitigate or fully inform ADRD, which is an injustice. However, including them presents unique challenges, as their impairment often means they have a diminished capacity to give consent, which is necessary for participation. To ensure their representation, it is important to consider the ethical implications of their inability to consent. For those with diminished capacity, a substitute decision-maker, formally appointed guardian, or family carer can act on their behalf [26]. This is particularly important for microbiome studies, which can be long-term and require ongoing engagement, reaffirmation of consent, and acknowledgment of new information as it becomes available [27].

For individuals who have historically faced discrimination, the intersectionality of ADRD and cultural/racial challenges can be significant. Researchers must be culturally competent with the population of interest and understand how various cultures approach decreased cognition and participation in research. This is critical for advancing science in ways that benefit the broader population. For example, the lived experiences of people with ADRD vary across cultures and belief systems. Building relationships with community workers who share the cultural and racial backgrounds of the targeted community is essential; this may include a bilingual community coordinator. Researchers should also collaborate with trusted community-based organizations, such as community centers, local Alzheimer’s resource centers, churches, and other faith communities. Previous research has shown the importance of church and faith in Black older adults, and recruiting from healthcare facilities may not be ideal for minority older adults due to lack of healthcare access due to limited insurance, transportation, and financial resources [28]. Although electronic methods of engagement are becoming more common, in-person, interactive recruitment opportunities may be more effective where trust is a challenge. Additionally, researchers must provide comprehensive support for ADRD to under-represented populations beyond access to research-related information [29]

Developing exclusion criteria

While much progress has been made in the identification of patients with ADRD using clinical, radiographic, and laboratory-based testing, identifying potential exclusion criteria is important to consider in the design of microbiota clinical trials involving patients with cognitive impairment. Older adults with impaired gastrointestinal function (Table 1) may impact the person’s microbiota. Other exclusion criteria are needed to identify participants with recent illnesses that will directly impact gastrointestinal microbiota. The use of recent antibiotics and recent diarrheal illness are two such examples. With regards to antibiotics, participants using ocular solution antibiotics should not be excluded. On a physical exam, patients with severe pedal or sacral edema often have edema of the colon, which could significantly affect gastrointestinal absorption and function.

Table 1.

Potential exclusion criteria related to gastrointestinal dysfunction

Intravenous, intramuscular, or oral antibiotics within the previous 6 weeks Inflammatory bowel disease, i.e., Crohn’s disease or ulcerative colitis (ICD- 10 K50.0X, K50.90, or K51.9) Diagnosis of celiac disease (ICD- 10 K90.0), non-celiac gluten intolerance (ICD- 10 K90.41); exocrine pancreatic insufficiency (ICD- 10 K86.81), dermatitis herpetiformis (ICD- 10 L13.0) Percutaneous endoscopic gastrostomy (PEG) or percutaneous endoscopic jejunostomy (PEJ) Colostomy (ICD- 10 Z93.3) Current diagnosis of generalized edema or anasarca (ICD- 10 R60.1) or fluid overload (ICD- 10 E87.70) Recent diarrheal illness (within the last 2 weeks)

Since microbiome interventions will likely need long-term follow-up, additional criteria are needed to exclude patients with a poor long-term prognosis (Table 2). For example, patients enrolled in a hospice program have a survival prognosis that is estimated to be less than 6 months. Similarly, patients with an end-stage illness often have a poor long-term prognosis that would limit their participation in a longitudinal study. Patients with severe unintentional weight loss may have a poor prognosis or a markedly decreased intake of food. Utilizing ICD- 10 coding for end-stage illnesses may allow for more efficient prescreening efforts in settings involving large healthcare systems.

Table 2.

Potential exclusion criteria related to advanced confounding illness

Enrollment in a hospice program Encounter for palliative care (ICD- 10 Z51.5) One or more end-stage medical illnesses (ICD- 10) that would interfere with longitudinal participation [23]—chronic kidney disease, stage 5 (N18.5); end-stage heart failure (I50.84); malignant neoplasm of the bronchia and lung (C34.X); malignant pleural effusion (J91.0); secondary malignant neoplasms (C78.X); secondary malignant neoplasm of the bone (C79.5X); malignant neoplasm of the colon (C18.X); malignant neoplasm of the prostate (C61); malignant neoplasm of the sinuses (C31.X); vascular dementia severe (F01.CXX); unspecified dementia, severe (F03.CXX); dementia in other diseases classified elsewhere, severe (F02.CXX); hepatic failure (K72.00, K72.01, K72.10, K72.11, K72.90, K72.91); idiopathic pulmonary fibrosis (J84.112) with dependence on supplemental oxygen (Z99.91); chronic obstructive pulmonary disease, unspecified (J44.9) with dependence on supplemental oxygen (Z99.91); other specified chronic obstructive pulmonary disease (J44.89) with dependence on supplemental oxygen (Z99.91); and amyotrophic lateral sclerosis (G12.21) Diagnosis of abnormal weight loss (ICD- 10 R63.4), cachexia (ICD- 10 R64), unspecified severe-protein-calorie malnutrition (ICD- 10 E43), or adult failure to thrive (ICD- 10 Z51.5) and a body mass index < 19.9 (ICD- 10 Z68.1) Serum albumin level less than or equal to 2.5 g per deciliter

Identifying potential confounding factors

The comparison of the gut microbiome of patients with ADRD versus older adults without cognitive impairment needs to control for a multitude of potential variables. Compared with younger populations, older adults have more heterogeneity regarding comorbid illnesses, functional capacity, and medication usage. Among Medicare beneficiaries over the age of 65, 38% had four or more chronic medical illnesses [30]. Many older adults have three or more chronic conditions. Older adults are also heterogeneous regarding diet, social determinants, and environmental factors that differ geographically and culturally [31].

Impact of disease

Multiple diseases have the potential to impact microbiome studies involving older adults with cognitive impairment. While we think of illnesses, such as cancer, diabetes, and autoimmune disease, as categorical variables (i.e., groupings based on the presence or absence of the disease), these illnesses are heterogeneous continuous variables [31]. Some of the variables listed in Table 3 can be extracted from an electronic medical record of homebound older adults or the medication administration record at a long-term care facility. However, other variables, such as the diagnosis and treatment of cancer, are highly individualized and will need to be documented manually. The study staff will need to carefully document the specific cancer diagnose(s), cancer treatment(s), and abdominal surgeries.

Table 3.

Disease factors that may impact microbiome studies

One or more pneumococcal vaccines (PCV 13, PCV15, PCV20, PCV21, PPSV23) History of cancer or current malignant disease (excluding non-melanoma skin cancers) Cancer treatment modality(s) used with approximate dates (i.e., type of surgery, radiation therapy, chemotherapy) Prevalence of gastrointestinal side-effects, such as diarrhea and vomiting Listing of all abdominal surgeries History or current diagnoses of gastrointestinal reflux disease (GERD) or dyspepsia Current diagnosis of diabetic gastroparesis (ICD- 10 E11.43) or gastroparesis (ICD- 10 K31.84) Substance abuse disorder (ICD- 10 F19.10) Alcohol abuse/dependence/use ICD- 10 F10.XXX) Major depression (ICD- 10 F32.X, F33.X) Smoking status (non-smoker, smoker, former smoker, unknown)

Impact of medications and nutritional supplements

Because of the higher prevalence of chronic medical conditions, older adults use more medications on average than younger patients [32]. Many commonly used medications have a direct impact on gastric acidity (i.e., pH level). Some of these medications, which include common over-the-counter agents, block the release of gastric acid. Other medications, including acetylcholinesterase inhibitors (used in the treatment of patients with mild cognitive impairment and Alzheimer’s disease), stimulate gastric acid secretion. Gastric acid prevents the growth of microbiota in the stomach and small intestines and prevents bacteria from traversing from the oral cavity to the colon. Medications used to treat constipation will directly impact the function of the colon and thus affect the gut microbiota. Glucocorticoids and biological immunotherapy have a direct impact on the immune system and thus alter the potential organisms comprising gut microbiota.

For all drugs listed in Table 4, it is important to document the specific medication along with the dosage and frequency of use per day. Probiotics and supplements (Table 5) should list the specific brand name along with the dosage and frequency. In the case of probiotics, different brands contain different organisms that are intended to colonize the colon. As with the items listed in Table 3, the items in Tables 4 and 5 could be extracted from an electronic medical record of homebound older adults or the medication administration record at a long-term care facility.

Table 4.

Medications that may impact microbiome studies involving older adults with cognitive impairment

Hydroxymethylglutaryl-CoA reductase inhibitor (statins) medications containing atorvastatin, rosuvastatin, simvastatin, pravastatin, fluvastatin, or lovastatin Thyroid hormone supplements—levothyroxine Fibrates—clofibrate, fenofibrate, or gemfibrozil Anti-ulcer proton pump inhibitors medications containing omeprazole, pantoprazole, lansoprazole, esomeprazole, or rabeprazole Anti-ulcer histamine- 2 blocker medications containing cimetidine, famotidine, nizatidine, or ranitidine Anti-ulcer drug sucralfate Acetylcholinesterase inhibitor medications containing donepezil, galantamine, or rivastigmine Alpha-glucosidase inhibitor medications containing acarbose or miglitol GLP- 1 receptor agonist medications containing dulaglutide, exenatide, liraglutide, semaglutide, or tirzepatide Osmotic laxatives containing lactulose, polyethylene glycol, or magnesium citrate Stimulant laxatives containing bisacodyl and/or senna Docusate (stool softener) Current or recent (less than 6 months) use of biologics for cancer, rheumatoid disease, inflammatory bowel disease, or multiple sclerosis containing adalimumab, infliximab, etanercept, rituximab, etc Current use of oral glucocorticoid immunosuppressant medications containing prednisone, prednisolone, dexamethasone, and methylprednisolone. The use of ocular glucocorticoids will not be documented Current use of inhalation glucocorticoid medications containing budesonide, fluticasone, beclomethasone, mometasone, or ciclesonide

Table 5.

Supplements that may impact microbiome studies involving older adults with cognitive impairment

Probiotics supplements Oral nutritional supplements (i.e., Boost, Ensure, Nepro, Glucerna) Herbal or other nutritional supplements Listing of all vitamin supplements Listing of all mineral supplements

Linking microbiota data with specific dietary factors

Multiple survey formats exist that can provide much information about the nutritional components of common foods. Most of these surveys rely on participant or caregiver self-report. Recommended dietary assessment methods include but are not limited to food frequency questionnaires (FFQ), 24-h dietary recall survey platforms, or caregiver-assisted recall. Some questionnaires require long-term memory, which may be challenging with patients with cognitive impairment. The length of surveys needs to be limited for caregivers of research participants with ADRD, as many of whom are under severe emotional distress. In addition, these surveys require trained staff to administer. [33]

The Nutritional Data System for Research (NDSR) and Automated Self-Administered 24-h (ASA- 24®) [34] are two of the existing 24-h dietary recall survey platforms [35]. Twenty-four-hour dietary recall surveys rely on self-reported information and require the assistance of a trained staff member to administrate. In addition to measuring the quantity of food intake, other dietary factors should be documented (Table 6). Most of these factors would need to be obtained through an interview of the research participant or the caregiver [3, 36]. The recruitment of multiple older adult participants from the same institutionalized setting may represent an opportunity to control for dietary factors, including food preparation and food composition. Standardization techniques, like pre-study dietary run-ins, are recommended to control for confounding variables [37].

Table 6.

Dietary factors to include in microbiome recruitment studies

Body mass index Intake of artificial sweeteners Diet type (conventional, vegetarian, vegan, organic, carnivore, keto, etc.) Cultural diet Presence of difficulty swallowing (dysphagia) Need for assistance or supervision with feeding Need for assistance with food preparation

Implications and interactions of medications and supplements with microbiome

Medications and supplements can significantly influence the microbiome and should be systematically accounted for in data analysis, particularly in studies investigating aging and dementia. The human gut microbiome is highly responsive to external factors, and pharmacological agents and dietary supplements can cause profound shifts in microbial diversity, composition, and function. This has critical implications in microbiome research, where failing to account for these variables may lead to confounding results and misinterpretation of associations between the microbiome and health outcomes. In dementia research specifically, common pharmacologic treatments such as cholinesterase inhibitors (e.g., donepezil, rivastigmine, galantamine) and NMDA receptor antagonists like memantine are frequently prescribed to manage cognitive symptoms. These agents can indirectly or directly impact gut microbiota via altered gastrointestinal motility, secretion, and immune interactions. For instance, cholinesterase inhibitors are known to increase acetylcholine levels in the gut, which can modulate gut motility and potentially change microbial fermentation dynamics and transit time—two critical factors influencing microbiota structure. Memantine, although primarily acting on central nervous system receptors, may also influence the gut-brain axis pathways, thereby altering microbiome-related metabolic signaling.

Similarly, widely used dietary supplements such as probiotics, prebiotics, omega- 3 fatty acids, and multivitamins are known modulators of gut microbiota. Probiotics directly introduce live microbes into the gut ecosystem, which can transiently or persistently alter microbial community composition depending on strain-specific colonization potential. Omega- 3 fatty acids have been shown to influence microbial diversity and enhance the abundance of anti-inflammatory taxa such as Lactobacillus and Bifidobacterium. These shifts can potentially mimic or mask the microbiome signatures associated with cognitive function or neurodegenerative disease progression, thereby confounding the results of microbiome-dementia association studies if not properly accounted for.

To address these complexities, researchers should incorporate detailed data collection on medication and supplement use at baseline and throughout the study period. This includes information on dose, duration, frequency, and any changes over time. In a statistical analysis, these factors should be included as covariates in multivariable models to control for their confounding effects. Advanced modeling approaches such as mixed-effects models, propensity score matching, or Bayesian hierarchical models can further refine adjustments by accounting for time-varying exposures and individual-level heterogeneity. Where feasible, stratified analyses can be conducted to examine microbiome associations within specific medication or supplement user subgroups. Another valuable strategy is to perform sensitivity analyses excluding individuals with high-impact exposures (e.g., probiotic users or individuals on multiple medications) to assess the robustness of primary associations.

From a study design perspective, researchers should consider matching participants on medication/supplement use or utilizing randomization in interventional trials to mitigate confounding. Longitudinal study designs are particularly beneficial in capturing dynamic changes in both the microbiome and exposure profiles, enabling temporal causal inference. Moreover, integrating pharmacomicrobiomics analyses—examining how microbiota modulate drug metabolism and vice versa—can uncover novel insights into host-microbiome-drug interactions relevant to dementia outcomes. Ultimately, systematic consideration and methodological rigor in handling medication and supplement influences are essential to accurately disentangle microbiome associations from external modifiers and to advance precision microbiome research in aging and dementia.

Conclusion

While prior ADRD microbiome studies account for differences in cognitive testing scores, the gut microbiome is affected by multiple complex medical, physiological, pharmacological, and social and dietary factors. Many of these variables could impact the external validity and findings of prospective microbiome studies. The framework described in this manuscript will allow for the recruitment and analysis of participants that reflect the diversity of the cognitively impaired older adult population. Older adults are a diverse segment of the population in terms of age, race, ethnicity, and functional impairment. Without significant attention to recruiting diverse populations in microbiome studies, health inequities in ADRD may worsen.

Author contribution

A.G. conceived the presented idea and wrote the manuscript with support from C.W., M.M., H.Y., and C.L. A.A., P.H., and S.J. edited the manuscript. All authors approved the final version for submission.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

All authors consent to the final version of the manuscript for publication.

Conflict of interest

Dr. Hariom Yadav is a co-founder and the Chief Scientific Officer of Postbiotics Inc. He is also a co-founder of BiomAge Inc., MusB LLC, and MusB Research LLC, alongside Dr. Shalini Jain. The authors declare that there are no conflicts of interest related to the studies and findings presented in this manuscript.

Disclaimer

The contents of this publication do not represent the views of the Department of Veterans Affairs or the United States Government.