High Prevalence, Different Decisions: Dementia, Thyroid, Diabetes, and Sleep Apnea
Adults with Down syndrome face several conditions that deserve close attention, but the evidence does not turn them into one interchangeable screening problem. The important comparison is how age pattern, symptom visibility, test burden, treatability, and downstream evidence shape the form of attention. Dementia illustrates longitudinal surveillance; thyroid disease and diabetes illustrate weak-for laboratory recommendations despite absent outcome trials; sleep apnea illustrates high clinical concern with widely varying prevalence estimates. Together, these cases show why a high-prevalence condition can justify vigilance without automatically justifying the same test, interval, or claim of benefit.
Dementia and behavioral change require a baseline, not just an age label
Dementia is the clearest example of why the question “Should this person be screened?” may be less useful than “Has this person changed from their established baseline, and what could explain the change?” Alzheimer’s disease is substantially more common among adults with Down syndrome. In a large U.S. retrospective cohort, Alzheimer’s disease was coded in 12.16% of adults with Down syndrome compared with 0.38% of age- and sex-matched controls. The comparison is observational and depends on diagnostic coding, but the direction of the difference is consistent with the age-related pattern described in other sources.
Age is essential to interpreting that pattern. The AHRQ review states that clinical dementia rarely occurs before age 40, while by age 60 at least one-third and possibly more than two-thirds of people with Down syndrome may have Alzheimer-type dementia. The range itself is a warning against treating a single percentage as a universal forecast: the estimate depends on how dementia was defined, who was studied, and how participants were assessed. The age relationship is more stable than any one number. Risk rises markedly with age, so an assessment designed to detect change in midlife is not equivalent to labeling a young adult with a neurodegenerative disease.
The 2020 GLOBAL guideline responds with two linked recommendations. It urges caution when diagnosing age-related Alzheimer-type dementia in adults younger than 40 because its prevalence is low before that age. It also strongly recommends annual assessment beginning at age 40 of changes from baseline, with an interview of primary caregivers and attention to cognition, memory, executive function, behavior, personality, communication, adaptive functioning, ambulation, motor skills, and established abilities. The difference between these recommendations is instructive. Caution about a diagnosis is not a recommendation to stop observing. Active observation is not the same as assuming that any new difficulty proves dementia.
A baseline changes the information available to the clinician. A one-time cognitive score asks how a person performs relative to a test or a reference group. A longitudinal history asks whether a person who usually communicates, walks, manages routines, or participates in familiar activities is doing something materially different. For adults with Down syndrome, where expressive language and intellectual functioning vary widely, a change across several functional domains may be more informative than a generic comparison with people without Down syndrome. A caregiver who knows the adult’s ordinary routines can identify a gradual loss of skills that a brief encounter might miss. The adult’s own account remains important, but the guideline treats knowledgeable observers as an additional source of evidence rather than assuming that self-report alone captures every change.
The baseline also protects against diagnostic overshadowing. New slowing, withdrawal, loss of initiative, disrupted sleep, or difficulty with daily tasks can be compatible with dementia, but they can also reflect hypothyroidism, obstructive sleep apnea, seizures, chronic pain, medication effects, sensory impairment, metabolic disturbance, or a mental-health condition. If the first explanation is “this is Down syndrome” or “this must be Alzheimer’s disease,” the process can fail in opposite directions. A genuine decline may be overlooked, or a treatable problem may be given an irreversible label. The guideline’s reasoning is that finding a reversible contributor can have value even when the final diagnosis is not dementia.
Behavioral change makes this differential diagnosis still more important. The AHRQ review describes behavioral and mental-health conditions as difficult to disentangle from baseline traits, communication differences, cognitive impairment, medical comorbidity, and aging. In one large U.S. coded cohort, depression and obsessive-compulsive disorder were more common among adults with Down syndrome than matched controls, but those administrative comparisons do not establish that symptoms present in the same way, or that a particular behavioral questionnaire improves diagnosis. Indeed, the review found no eligible studies directly comparing the clinical symptoms and presentation of common behavioral or mental-health conditions in adults with Down syndrome with those in the general adult population.
That missing comparison matters because the meaning of a symptom depends on context. A person who takes much longer to complete a familiar routine may be experiencing depression, sleep-related impairment, pain, medication effects, or cognitive decline. A person who appears irritable or disengaged may be reacting to hearing or vision loss. A sudden loss of language and self-care in a younger adult may require a specialized evaluation rather than an assumption of ordinary aging. The available evidence does not provide a simple algorithm that assigns each behavior to one diagnosis. It supports a broader clinical question: what changed, when did it change, what other systems could produce it, and which next assessment could distinguish those possibilities?
Treatment evidence does not remove this uncertainty. Dementia has the largest treatment literature in the AHRQ review, including exercise, donepezil, memantine, supplements, and a vaccine. Some exercise studies suggested reduced risk of decline in selected domains, and isolated drug findings were favorable, but most cognitive outcomes showed no significant difference or could not be combined. The review rated the evidence insufficient because of small samples, inconsistent measures, unknown consistency, imprecision, and study limitations. Those treatment findings cannot be used as proof that detecting dementia earlier through routine screening improves outcomes.
The practical implication is a two-part discipline. First, do not use low prevalence before 40 as a reason to ignore a real change or a medical cause of that change. Second, do not use high prevalence after 40 as a reason to turn every change into an Alzheimer-type diagnosis. A longitudinal baseline-and-change approach creates a way to be vigilant without making the label do more work than the evidence allows.
The diagram’s most important feature is not the label at the end. It is the return to an updated baseline. A person’s health status changes, and the meaning of later observations depends on what has already been learned. This is why an annual conversation can be clinically substantive even when it does not produce a laboratory value or a definitive diagnosis.
Thyroid disease, diabetes, and sleep apnea show different ways to act on concern
Thyroid disease and diabetes are useful contrasts with dementia because the GLOBAL guideline supports periodic laboratory screening, but with weak-for recommendations. The guideline recommends serum thyroid-stimulating hormone testing for hypothyroidism every one to two years beginning at age 21. It recommends hemoglobin A1c or fasting plasma glucose for type 2 diabetes every three years beginning at age 30, and every two to three years beginning at age 21 for an adult with Down syndrome and obesity. These are concrete recommendations, but their strength and rationale must be kept visible: neither condition had direct evidence showing that screening asymptomatic adults with Down syndrome improves clinical outcomes.
Why recommend testing at all? For hypothyroidism, the problem is both prevalence and symptom ambiguity. The guideline cites several studies that found hypothyroidism in 39% to 61% of adults with Down syndrome, while the AHRQ review reports much lower estimates in some administrative datasets and much higher estimates in other cohorts. One U.S. record-based comparison reported 36.4% in adults with Down syndrome versus 5% in controls; a U.K. case-control study reported 22.9% versus 1.8%; a French national database produced a much lower overall estimate. These numbers are not a tidy disagreement about one true rate. They reflect different diagnostic practices, populations, and ways of identifying disease.
The clinical reasoning does not depend on selecting the largest percentage. Symptoms commonly associated with hypothyroidism—fatigue, weight gain, constipation, and mental or functional changes—may also occur for other reasons in adults with Down syndrome. Communication differences can make fatigue or temperature intolerance difficult to identify. If clinicians wait for a distinctive symptom pattern, they may miss a treatable condition. A blood test is relatively familiar and the guideline regards monitored thyroid-hormone treatment as generally safe and inexpensive. Those considerations create a plausible benefit pathway: laboratory testing may identify a condition that can be treated before nonspecific changes become more difficult to interpret.
But plausibility is still not a screening-outcome trial. The guideline explicitly states that no study had shown that TSH or T4 screening improved clinical outcomes in adults with Down syndrome. It also identifies burdens: phlebotomy can cause anxiety and take time, and treatment can be inappropriate or excessive if an abnormal result is interpreted without attention to the individual. The recommendation is weak for because the workgroup judged potential benefits to outweigh potential harms, not because the optimal interval has been directly measured. The AHRQ review’s later finding of no screening studies reinforces the difference between a reasoned recommendation and a demonstrated net effect.
The diabetes recommendation follows a related but not identical path. Obesity is common among adults with Down syndrome, and the guideline treats it as a risk factor for type 2 diabetes. It also points to potential long-term complications of uncontrolled diabetes, including damage involving the eyes, nerves, kidneys, and cardiovascular system. A large U.K. study cited by the guideline found higher diabetes prevalence in adults with Down syndrome than in matched controls, although it did not distinguish type 1 from type 2. The more recent AHRQ review reports that type 2 diabetes incidence was higher in people with Down syndrome up to age 44, similar between groups from 45 to 54, and lower in the Down syndrome group at older ages in one population-based analysis. That pattern is a reminder that “higher risk” can depend on age and outcome, not only on diagnosis.
The guideline therefore adapts general-population reasoning rather than claiming a Down syndrome-specific trial has proven the intervals. It recommends earlier screening because of premature aging, obesity, the potential for earlier complications, and the possibility that impaired organ systems may make complications consequential. It also recognizes the harm pathway: a blood draw is a burden, and a diagnosis can lead to overly strict dietary rules, nonbeneficial medication, hypoglycemia, or side effects. The decision is defensible only when the downstream care is individualized and the person, family, and caregivers understand what the result may change.
Sleep apnea presents a third pattern. The condition can cause fragmented sleep, reduced oxygen levels, daytime fatigue, and behavioral or cognitive effects, and it is associated with cardiovascular and neurocognitive problems. Yet prevalence estimates in adults with Down syndrome vary dramatically according to ascertainment. In U.S. electronic health records, one comparison reported obstructive sleep apnea in 9.56% of adults with Down syndrome versus 5.65% of controls. In a U.K. study of adults not already treated for sleep apnea, 35% screened as probable cases by questionnaire, and 42% of the smaller group who agreed to polysomnography had sleep apnea. Studies of selected adults referred for sleep testing reported still higher rates, including 78.7% or 100% in some samples.
Those figures should not be averaged into a single universal prevalence. A medical record records recognized diagnoses. A questionnaire identifies people with a pattern of reported symptoms. A sleep study measures physiology in the people who complete it. A referred sample has already passed through a clinical suspicion or access filter. Each method answers a different question, and refusal or inability to complete a sleep test can further affect who is represented.
The treatment evidence is encouraging in places but remains insufficient. A small randomized CPAP pilot in adults diagnosed with obstructive sleep apnea found no significant difference in sleepiness at one month, while longer follow-up was associated with improvements in self-reported sleepiness, disruptive behavior, anxiety or antisocial traits, depressive symptoms, and verbal and nonverbal cognitive scores. Adaptive skills and health-related quality of life did not significantly improve. Small surgical studies in adults with moderate to severe sleep apnea who could not tolerate positive airway pressure reported improvements in apnea severity and oxygenation, but both were highly selected, single-arm studies with critical risk of bias. These results support the importance of treatment research; they do not establish a universal screening interval or prove that finding every asymptomatic case produces better outcomes.
The shared lesson across the four conditions is not that one approach always wins. Dementia benefits from a longitudinal baseline because the important signal is change across function and because premature labeling can hide reversible causes. Thyroid and diabetes receive weak-for laboratory recommendations because high concern, nonspecific symptoms, treatability, and manageable testing burdens may favor earlier detection, even though direct screening benefits remain unmeasured. Sleep apnea warrants serious attention because the condition and its treatment can matter, but ascertainment and treatment evidence do not yet support treating every prevalence estimate as a universal screening mandate.
High prevalence supplies a reason to ask a more precise question. It does not supply the answer in advance. The next cases turn to situations in which a test, image, or transferred risk rule may create restriction or overtreatment when its connection to the patient-important outcome is poorly calibrated.