I lost the word "culvert" in a progress meeting last winter. I could see the thing. I could describe what it did and where it sat on the drawing. The word turned up about ninety seconds later, by which point somebody had finished the sentence for me and we had moved on to the next item.

Nobody said anything. What stayed with me was the tally. I had been keeping a quiet private one of these for a while, and it was longer than I would have guessed.

Before the rest of it: I am not a clinician and none of this is a diagnosis. The next step for a man who has noticed this in himself is a GP appointment, and that is a smaller thing than it looks from here. What I can offer is the reading, ordered by how much weight the evidence carries. The order surprised me. The explanation men reach for first has the least behind it.

How common it is, and how rarely it gets raised

Roughly one man in nine at this age reports it, and fewer than half of the people reporting it have told a doctor.

The US Centers for Disease Control published the numbers in MMWR in July 2018, drawn from the 2015-2016 Behavioral Risk Factor Surveillance System across 49 states, DC and Puerto Rico. It measured what the researchers call subjective cognitive decline: self-reported worsening or more frequent confusion or memory loss over the past twelve months.

At 45 to 54 it was 10.4 per cent. At 55 to 64, 11.4 per cent. Across men, 11.4 per cent. Only 45.4 per cent had discussed it with a health care provider.

The figure I keep coming back to is a different one. Among people reporting subjective cognitive decline who also reported functional difficulties, 73.3 per cent said it interfered with work or social activities. A man who cannot follow the thread in a noisy pub stops going to the pub, and the withdrawal reads to everyone else, including him, as ordinary midlife busyness.

That is US survey data resting on self-report, which is a real limitation. It tells you the experience is common and under-reported, and nothing at all about what is causing yours.

Sleep carries the most weight, and it is still not settled

The single best-evidenced association is between short sleep in midlife and later cognitive decline.

Séverine Sabia and colleagues published the analysis in Nature Communications in April 2021, using Whitehall II, a British civil-service cohort that is roughly two thirds male: 7,959 participants followed from age 50, a mean follow-up of 24.6 years, and 521 recorded dementia cases. The figures below are for the whole sample, men and women together, because that is how the paper reports them. Sleeping six hours or less against seven gave a hazard ratio of 1.22 at age 50, 1.37 at 60 and 1.24 at 70. Participants who were persistently short sleepers across all three points sat at 1.30.

Now the honest part, because this study gets quoted far more confidently than it deserves. The confidence intervals cross or touch 1.00 in two of those four estimates (1.01-1.48, 1.10-1.72, 0.98-1.57, 1.00-1.69), which means two of them are compatible with no effect at all. The authors themselves flag residual confounding and reverse causation, and reverse causation matters here: dementia has a long preclinical period, so poor sleep at 60 may be an early symptom rather than a cause of what shows up at 75.

None of that makes it settled. It supports sleep being the most promising thing on this list and the cheapest to look at, and it does not support a claim that fixing your sleep at 50 prevents anything.

Sleep apnoea is the one I would rule out first

If you take one action out of this article, get assessed for obstructive sleep apnoea. It is common in men, testable and treatable.

Karl Franklin and Eva Lindberg reviewed eleven population studies published between 1993 and 2013 in the Journal of Thoracic Disease in August 2015. Obstructive sleep apnoea at an apnoea-hypopnoea index of 5 or above had a mean prevalence in men of 22 per cent, range 9 to 37 per cent, with the studies from 2008 to 2013 reporting 37 per cent.

Symptomatic apnoea, meaning the breathing events plus daytime sleepiness, came in at 6 per cent of men, range 3 to 18 per cent.

The gap between those two numbers is the useful bit: only a fraction of population cases have the daytime sleepiness that makes anyone suspect it, so a large share go clinically unrecognised. The review I am citing counts cases. It does not test anybody's memory, so the reason I would start here is mine and not its finding: apnoea is common, testable and treatable, which is a better combination than most things on this list. A man can have significant apnoea, not feel especially sleepy, and simply notice that his head does not work as well as it did.

Alcohol shows a dose-response, and no protective floor

Alcohol has the clearest gradient of anything here, and the light-drinking exemption most of us assumed we had does not survive it.

Anya Topiwala and colleagues published in the BMJ in June 2017: 527 Whitehall II participants, mean age 43 at baseline, drinking measured repeatedly over 30 years, with MRI at the end. Against abstainers, the odds of hippocampal atrophy were 3.4 at 14 to 21 units a week (1.4-8.1, p=0.007) and 5.8 above 30 units (1.8-18.6). Higher intake also predicted faster decline in lexical fluency, though not in semantic fluency or word recall.

There was no protective effect of light drinking over abstinence. People on one to under seven units a week did no better than people drinking nothing.

Two things to keep straight. This is observational, so it shows association rather than proof of cause, on a sample of 527. And the units are British: a UK unit is 8 grams of alcohol and an Australian standard drink is 10, so that 14 to 21 unit band is roughly 11 to 17 standard drinks a week, and 30 units is about 24. Two glasses of wine most nights puts a man in the middle of that band without any sense of drinking heavily.

For scale in Australia, the AIHW National Drug Strategy Household Survey 2022-23 found 39 per cent of males aged 14 and over drinking at risky levels. That is everyone from 14 up, and the AIHW page gives no breakdown for men in their forties and fifties, so I am not going to imply one.

Thyroid: a common suspicion with a trial against it

Underactive thyroid is worth ruling out with a blood test, but treating a borderline result is unlikely to do anything for your head.

The TRUST trial (Stott and colleagues, New England Journal of Medicine, 29 June 2017) was double-blind, randomised and placebo-controlled: 737 adults with persistent subclinical hypothyroidism, mean age 74.4. Levothyroxine worked biochemically, pulling TSH from 6.40 down to 3.63 against 5.48 on placebo. At one year the difference in the Hypothyroid Symptoms score was 0.0 (-2.0 to 2.1) and in the Tiredness score 0.4 (-2.1 to 2.9), against a minimum clinically important difference of nine points. There was no benefit on the secondary measures either, including executive cognitive function.

The boundary matters: inclusion was 65 and over, so this does not directly test 50-year-olds. It does take a lot of the optimism out of a treatment that gets prescribed hopefully.

Hormones are the weakest leg of the argument

This is the explanation the internet offers first, and it is the one with the least support.

The Testosterone Trials (Snyder and colleagues, Endocrine Reviews, March 2018) enrolled 788 men, mean age 72, for a year across seven coordinated trials. The Cognitive Function Trial was one of them and ran on the subset with age-associated memory impairment rather than on all 788. It found no improvement in delayed paragraph recall, visual memory, spatial ability, executive function, subjective memory complaints or global cognition.

The American College of Physicians guideline (Qaseem and colleagues, Annals of Internal Medicine, 21 January 2020) puts it directly: do not initiate testosterone to improve energy, vitality, physical function or cognition. That is a conditional recommendation on low-certainty evidence, which is the honest strength of it, and it is still two guideline-level sources pointing the same way.

I am not saying hormones are irrelevant to how a man feels at fifty. I am saying that if the specific complaint is fog, the trial that tested exactly that found nothing, and the men in it were older than us.

What I would take to a GP

The appointment goes better with three things written down.

A fortnight of sleep, roughly. Time in bed, time actually asleep if you have any idea, and how many times you surfaced. Approximate is fine.

Whether you snore or stop breathing. You cannot answer this yourself. Ask the person next to you, and take their answer rather than your version of it.

Two weeks of honest drinking, in standard drinks. The actual fortnight, counted as it happens rather than reconstructed afterwards.

Then ask two questions plainly: is a sleep study warranted, and what routine bloods would be sensible here. You do not need to have worked out what it is before you go. That is the doctor's job.

The reason I wrote this up at all is that 73.3 per cent figure, and it is worth restating with its condition attached: among people who reported both subjective cognitive decline and functional difficulties, that share said it interfered with work or social activities. Fog that makes conversation hard produces withdrawal, and the withdrawal looks exactly like the ordinary drift I write about the rest of the time. If that is the half you recognise, what it feels like from inside is here and the version where the social life simply disappeared is here.

Sources