We analyzed recent trials of pharmacological treatments for Alzheimer’s disease and found that the age of the participants does not reflect the actual distribution of patients in the general population. Individuals older than 84 years contribute greatly to the number of people with Alzheimer’s disease (currently 38 %), a proportion that is expected to grow further in the coming decades (to about 51 % in 2050) [14]. Only 8 out of 100 trial participants in the sample fall into this class. In contrast, 15 % of patients enrolled were younger than 65 years, so they cannot even be defined as old. People younger than 80 are too widely represented in clinical trials, therefore, despite the fact that those aged 80 and more are the large majority with Alzheimer’s disease (72 %).
Involving old persons in clinical research is often considered too challenging because coexisting diseases, functional disability, and multi-drug treatments are common and may confound trial results. Some authors suggest that the differences between participants in observational and clinical studies and target population are magnified when research involves older adults and that this may reduce the generalizability [29]. Moreover, ethical issues linked to the inclusion of a vulnerable population have to be considered. However, any real or alleged obstacle to research participation should be weighed against the fact that excluding patients who represent the largest part of the population with the disease may lead to misinterpretation in the conclusions about the underlying neuropathology, diagnostic methods, or therapy [11].
In the case of Alzheimer’s disease and other conditions that overwhelmingly affect old people, the inclusion of representative samples of patients in trials testing innovative pharmacological approaches and new drugs should be the rule. Older people have different pharmacokinetics and pharmacodynamics from people in their 60s or 70s [30]. These diversities may translate into different efficacy and safety profiles; benefits as well as harms may be under- or overestimated, depending on the disease expression, and must be assessed before new drugs are used. The results of robust and valid trials (internal validity) are also only clinically useful if they are relevant to a definable patient population that should represent the target group of patients for the intervention under study. This concept, known as external validity, applicability, or generalizability, is often neglected, especially in industry-sponsored research [10], which selects the best possible experimental conditions to highlight the efficacy of new treatments, often disregarding their effectiveness in the real world.
Trial participants should reflect the actual distribution of individuals with Alzheimer’s disease in terms of other important aspects too, such as co-morbidities, cultural and educational background, and frailty. These variables may threaten the external validity of clinical trials in Alzheimer’s disease as well as age.
In clinical practice too, subjects treated with the marketed drugs are a small proportion of those with Alzheimer’s disease. For instance, the largest proportions of patients treated with acetylcholinesterase inhibitors were in the younger age groups, with a steep drop with age: from 55 % at 60–69 years to 19 % at 80–84 years [31]. This may be because of perceived low effectiveness of these drugs by both the caregiver and the physician, difficulties with diagnoses of dementia, significant concurrent pathologic abnormalities, adverse drug reactions, or even a fatalistic acceptance of the condition. It is unlikely, however, that the reasons behind these therapeutic decisions are linked to strict application of evidence generated in a younger population.
Our analysis has several advantages. It focuses on the most recent research on Alzheimer’s disease and adds to previous reviews of trials of older interventions [8, 11]. We retrieved published and unpublished trials on putative disease-modifying agents for Alzheimer’s disease using a systematic approach and based our analysis on a large sample of studies, covering different pharmacological treatments, mechanisms of action, trial sponsors, and countries. The sample can be considered highly representative of current research in the field of Alzheimer’s disease, even if we could not retrieve reports of about one-third of the studies included. While it is uncertain whether the inclusion of these trials would have changed our findings, it is worrying that so many trial reports are still inaccessible.
One further limitation is the lack of reporting on the proportions of patients per age class in clinical trials. The age of the study population is generally described using mean and standard deviation or median and range. Thus, we could only estimate the proportion of patients in each age class assuming a specific underlying distribution, i.e., taking a normal distribution for reference. This approximation, however, is likely to have had only a limited impact on the overall conclusion of the review, given the very large age differences between trial participants and patients in the general population. To understand the applicability of the evidence from clinical trials to oldest people better, complete reporting of the age distribution of the patients is a major, necessary step forward.
Several studies reported prevalence data on dementia [32], while a few focused specifically on Alzheimer’s disease. To estimate the number of people with Alzheimer’s disease by age class in the general population, we used prevalence data from different sources. This is because the most recent figures we found indicated only the number of people with Alzheimer’s disease by 10 year classes of age [14]. Thus, to retrieve Alzheimer’s disease prevalence estimates by 5 year classes, we referred to published and unpublished data from previous cohort studies [17, 18].
Finally, we compared the ages of participants in clinical trials only with the number of people with Alzheimer’s disease in the USA because more than half the trials involved clinical centers in North America. Comparison with data from other population studies could be of interest. However, our estimates are likely to be conservative as taking as reference the populations of European countries, Australia, Canada, and Japan, where the proportions of older subjects (older than 65) and oldest subjects (older than 80 or 85) are even larger than in the USA, would have led to a more dramatic difference.