Why the same patient can be impaired on one test and normal on another
Cognitive screening tests are not interchangeable yardsticks. The MMSE and MoCA both produce a single global score, but they emphasize different cognitive domains and use different cutoffs. In a large Chinese community sample of adults aged 55 and older, the overall rate of mild cognitive impairment was 28.6% by MMSE and 36.2% by MoCA [2]. The two tests correlated well, but their agreement on who had impairment was only moderate, meaning a substantial number of people were classified differently depending on which test was used [2].
The MoCA was less prone to a ceiling effect, meaning it better separated people whose cognition was subtly below normal from those who were truly high-functioning [2]. That matters in sleep apnea, where the typical profile is not global dementia but selective weaknesses in attention, executive function, memory, processing speed, and visuospatial ability, while language and immediate recall are relatively preserved [3]. A test that samples those vulnerable domains more heavily will flag more people.
The size of the gap in sleep apnea studies
The meta-analysis makes the test effect concrete. Across studies of adults with objectively confirmed obstructive sleep apnea, the pooled impairment rate was 36.92%, but the rate varied significantly by assessment tool: 42.20% with the MoCA, 17.51% with the MMSE, 54.24% with the CDR, and 27.66% with the Critical Flicker Frequency test [1]. The CDR is a staging instrument originally designed to rate dementia severity, so it is not surprising that it labels more people as impaired than a brief global screen like the MMSE.
These differences are not just statistical noise. The confidence intervals for the MoCA and MMSE subgroups do not overlap, which supports the interpretation that the tool itself shifts the measured prevalence [1]. In other words, part of the wide range in published rates is a measurement artifact: studies using different tests are partly answering different questions.
Severity matters, but the test still changes the picture
Sleep apnea severity also moves the numbers. The meta-analysis found impairment rates of 32.22% in mild OSA, 36.79% in moderate OSA, and 44.46% in severe OSA [1]. A separate prospective study using the MoCA found a similar gradient: median MoCA scores were 28.5 in people without significant OSA, 26.5 in mild OSA, 25.0 in moderate OSA, and 23.5 in severe OSA, with lower scores indicating worse cognition [5]. That study also reported a negative correlation between MoCA score and the apnea-hypopnea index, the standard measure of breathing pauses per hour [5].
But severity and test choice interact. A test that is insensitive to the domains most affected by sleep apnea, such as the MMSE, may undercount impairment even in severe cases, while a more demanding or domain-specific instrument will detect more. The meta-analysis's severity subgroups were not stratified by test, so the two effects are tangled in the pooled numbers [1]. What is clear is that neither factor alone explains the range; both the patient's apnea severity and the instrument's sensitivity shape the reported rate.
What to keep in mind when you see a cognitive impairment rate
A reported rate of cognitive impairment in sleep apnea is only meaningful alongside the test used and the severity of the population studied. The meta-analysis also found higher rates in studies using full overnight polysomnography (38.61%) than in those using home sleep apnea testing (26.85%), and wide regional variation, from about 25% in the European Region to 62% in South-East Asia [1]. Those differences likely reflect case mix, diagnostic thresholds, and possibly cultural or educational effects on test performance, not just biology.
The broader point from the neurobiology literature is that cognitive assessment in sleep apnea should be domain-sensitive rather than based on a single global cutoff [3]. A single number from any one screening test is a rough summary, not a precise diagnosis. When comparing studies or reading a prevalence figure, the first question to ask is which test was used, because that choice alone can roughly double or halve the apparent rate.
Sources used in this answer
Prevalence of cognitive impairment among adults with obstructive sleep apnea: a systematic review and meta-analysis
In a meta-analysis of 23 studies, the pooled cognitive impairment prevalence in adults with obstructive sleep apnea was 36.92%, with significantly different rates by screening tool: 42.20% for MoCA, 17.51% for MMSE, 54.24% for CDR, and 27.66% for Critical Flicker Frequency.
A comparison of the Mini-Mental State Examination (MMSE) with the Montreal Cognitive Assessment (MoCA) for mild cognitive impairment screening in Chinese middle-aged and older population: a cross-sectional study
In a Chinese community sample aged 55 and older, mild cognitive impairment prevalence was 28.6% by MMSE and 36.2% by MoCA, with only moderate agreement between the two tests and less ceiling effect for MoCA.
Neurobiology of sleep apnea and cognitive impairment.
Obstructive sleep apnea impairs attention, executive function, memory, processing speed, and visuospatial ability while language and immediate recall are relatively preserved, and cognitive assessment should be domain-sensitive rather than based on a single global cutoff.
Obstructive sleep apnea is associated with cognitive impairment in minor ischemic stroke
In patients with minor ischemic stroke, moderate-to-severe obstructive sleep apnea was associated with worse MoCA scores and other cognitive measures compared with no or mild OSA, and MoCA scores were negatively correlated with apnea-hypopnea index and lowest oxygen saturation.
Assessing Cognitive Impairments in Obstructive Sleep Apnea Patients Using Montreal Cognitive Assessment (MoCA) Scores
In a prospective study of 72 participants, median MoCA scores declined with obstructive sleep apnea severity (28.5 in normal, 26.5 in mild, 25.0 in moderate, 23.5 in severe), and MoCA score was negatively correlated with apnea-hypopnea index.