TDG Programme Blog DH Concierge Book a Call
Clinical Practice

Why Your Last Health Plan Failed, And It Wasn’t Willpower

Two people get the same protocol. One follows it for eight months. The other abandons it in three weeks and concludes there is something wrong with them. The difference between them is measurable, it has been studied across more than a hundred thousand people, and it is not character.

STEPHEN DUNCAN FDN-P BSC HONS MSC · DETECTIVE HEALTH · AUGUST 2026
Well replicated · context-dependent · not a typology

After thirty-seven years I can predict, with uncomfortable accuracy and within about twenty minutes, whether someone will do what we agree.

I used to think that was clinical intuition. It is partly pattern recognition, and pattern recognition is unreliable in ways I have written about elsewhere. But it turns out the underlying thing is real, has been measured properly, and is more useful than the version in my head.

What The Evidence Actually Shows

Personality predicts adherence. Not vaguely — measurably, repeatedly, and at scale.

The Findings Worth Knowing

Bucher and colleagues, 2019, meta-analysed 99 studies covering 107,206 people. Lower neuroticism and higher extraversion, agreeableness, conscientiousness and openness were all associated with more favourable treatment outcomes. Agreeableness predicted the strength of the therapeutic relationship. Conscientiousness predicted follow-through.

Axelsson 2011, 749 people with chronic disease: neuroticism was negatively related to medication adherence, conscientiousness and agreeableness positively. Adachi 2022 put a number on it in cardiovascular patients — each point of conscientiousness raised the odds of good compliance by roughly 90%.

And the effects compound. Bucher found that treatment duration moderated the relationship: the longer the intervention, the more the traits mattered. Which is precisely the problem in my field, where nothing meaningful happens in under three months.

But Not Through Archetypes

Here is where I have to be careful, because there is an obvious and wrong next move.

Every practitioner develops a private taxonomy. Mine has categories in it. The rule-follower who executes a protocol flawlessly and comes apart when it changes. The senior manager who wants three things, will discard the fourth regardless of merit, and is a tougher nut who becomes a different person once you get past the exterior. The emotionally attuned client who needs to understand why before she will begin anything.

These are observations, not a validated system, and I am not going to present them as one. The history of personality typology in health is largely a history of frameworks that felt compelling and did not replicate — the cancer-prone personality being the best-known casualty. A taxonomy invented by one practitioner from one practice is a horoscope with clinical vocabulary.

What has been validated is the boring thing: five broad traits, measured with standard instruments, tested across large samples. That is what the studies above used. It is less charming than an archetype and it has the advantage of being true.

The Nuance That Makes It Useful

If the finding were simply "conscientious people comply," it would be true and useless. The interesting part is that the traits are context-dependent, and the context is often something you can change.

Jackson and colleagues examined 4,808 people and found the associations shift depending on circumstance. Neuroticism predicted non-adherence specifically when the treatment was perceived as expensive or as unimportant to health. Low conscientiousness predicted non-adherence specifically when people rated their own health as good — and when they thought the treatment was cheap and had side effects.

Read that again, because it inverts the usual framing. The anxious person does not fail generally. They fail when cost is salient. The disorganised person does not fail generally. They fail when they feel well.

Which means the intervention is not "be more conscientious." It is: make the cost feel proportionate for one, and make the stakes visible for the other.

Axelsson's follow-up work found something more actionable still. Self-efficacy — a person's belief that they can actually do the thing — mediated the relationship between traits and outcomes. Traits are broadly stable in adulthood. Self-efficacy is not. It responds to being given something achievable and then achieving it.

And Conscientiousness Itself Is Not Fixed

Jaworski and colleagues ran a twelve-week trial of a smartphone app combined with telephone coaching — values identification, specific goal-setting, tracking, scheduled calls — in people with multiple sclerosis and in healthy older adults.

Conscientiousness rose significantly in the intervention group, with a large effect size (d = 0.93), and no change in the waitlist controls. The effect was specific: no change in neuroticism, depression or anxiety. Small trial, and it needs replicating. But the trait most predictive of following a plan appears to be, to some degree, trainable.

The Finding I Found Uncomfortable

A 2024 nationally representative study of 1,003 people looked at two behaviours together: deliberately not following medical advice, and using alternative or complementary medicine.

Both were extremely common. And the strongest predictors of alternative medicine use were magical health beliefs and medical conspiracy beliefs, with low conscientiousness predicting deliberate non-adherence directly.

But the finding that stopped me was this one: negative experiences with the healthcare system facilitated both behaviours.

That describes a substantial part of my client base, and probably a substantial part of anyone's in this field. People arrive having been dismissed, rushed, or told their results were normal when they plainly did not feel normal. That experience is real and frequently justified. It also, on this evidence, makes them more likely to reject sound advice alongside the poor advice, and more receptive to explanations that are appealing rather than accurate.

Which puts an obligation on me rather than on them. If people arrive at functional medicine partly because conventional care let them down, then the last thing they need is a practitioner who exploits that opening with a more comfortable story.

It is one of the reasons I publish the things I have got wrong. Someone who has been sold certainty before deserves to see what it looks like when a practitioner does not have it.

What Any Of This Changes In Practice

Not the protocol. This matters and it gets confused constantly. Personality changes how a plan should be delivered, not what the plan should contain. The biochemistry does not care how you like to be spoken to.

What it legitimately changes
  • How many things are introduced at once
  • How much explanation comes before the first action
  • Whether progress is tracked visibly or lightly
  • How often we check in, and how
  • Whether cost is discussed openly and early
  • How much flexibility is built in from the start
What it must never change
  • What the test results actually mean
  • Whether a finding is reported honestly
  • Whether uncertainty is disclosed
  • Whether the answer is “you don’t need this”
  • Whether someone is referred onward

The rule-follower is the clearest example of both halves. Give them a plan and they will execute it exactly — which is an enormous asset, and the reason I like working with them. The vulnerability is that human physiology requires constant adjustment, and a plan presented as a rule becomes brittle the moment it needs to change.

So the adjustment is in the framing, not the content. Build the revision into the plan from the beginning. "We will review this at six weeks and change roughly a third of it" is not a hedge. For someone who needs rules, it makes revision itself the rule, and removes the sense that a change means something failed.

The Honest Limits

Three Things I Would Not Overstate

The evidence quality is uneven. A 2024 systematic review of personality and CPAP adherence found 13 of 21 studies were poor quality and only seven were high quality. The meta-analytic findings above are solid; much of the individual literature underneath them is not.

Most of it is cross-sectional and self-reported. Adherence measured by asking people whether they adhered. Traits measured by asking people what they are like. Both are subject to the obvious problems.

And I do not formally assess personality, nor am I suggesting you should be typed before you can be helped. I have not administered a Big Five inventory in my life. What the evidence does is tell me that something I was already responding to informally is real — which is a reason to do it deliberately rather than to build an instrument around it.

If You Are The One Whose Plans Keep Failing

The useful reframe is that repeated failure across different plans is information about fit, not about you. Five abandoned protocols is not five character failures. It is fairly strong evidence that the format was wrong every time, and nobody adjusted it.

Worth knowing about yourself, honestly:

Do you need to understand the mechanism before you will act, or does explanation delay you? Both are workable. Getting it backwards wastes months.

Do you follow a plan better when it is rigid or when it is loose? If rigid, you need the review dates built in. If loose, you need principles rather than protocols, and a detailed plan will feel like a cage.

What actually stopped you last time? Not "I lost motivation" — the specific thing. Cost, time, complexity, one component you disliked, or nothing visible happening for eight weeks. Each of those has a different fix and only one of them is about willpower.

And what has ever worked, even briefly? Usually there is something. The shape of the thing that worked tells you more than the list of things that did not.

None of this is a personality test. It is four questions, and the answers change how a plan should be built — which is a conversation worth having at the start rather than after the sixth attempt.

Educational content, not medical advice. Nothing here is a personality assessment or a psychological instrument, and the trait research described is population-level evidence, not a tool for categorising individuals.

Related

Why Your Last Health Plan Failed Your Thoughts Have a Blood Test The Abnormal Result That Might Be Protecting You More from the blog →

Test, don’t guess

Including about how you actually work. The best protocol is the one you are in a position to follow.

Ask AIdan → Book a discovery call