Step 7 of 8
When to Use Metabolic Interventions in Psychiatry
By the end of this lesson you will understand the specific clinical and biological patterns that make a metabolic approach worth considering in mental health care — and, just as importantly, the situations in which it is not appropriate.
Not every mental health difficulty is a fuel problem, and not every fuel problem announces itself clearly. The interesting question in metabolic psychiatry is not whether brain energy matters — it plainly does — but how a clinician decides that energy supply is the bottleneck worth addressing in a particular person. This lesson walks through the reasoning behind that decision, including the cases where the honest answer is no.
Imagine an office building where the lifts stall every afternoon, the servers reboot without warning, and the lights dim when the air conditioning kicks in. You could hire a new IT team. You could replace the lift motors. But if the underlying issue is that the building draws more power at peak demand than its supply can reliably deliver, each of those fixes will produce a partial, temporary improvement and then the same pattern will return. The question a good engineer asks is not what broke, but whether the failures cluster around moments of high demand.
This is a reasonable way to think about brain energy metabolism in psychiatry. The brain is roughly two per cent of body weight and consumes around twenty per cent of resting energy, most of it spent maintaining the electrical gradients that make signalling possible. That energy is produced in mitochondria, the cellular structures that convert fuel into usable currency. When that conversion is inefficient — because of mitochondrial dysfunction, chronic insulin resistance, or a loss of metabolic flexibility, the ability to switch smoothly between glucose and fat-derived fuels — the consequences are not evenly distributed. They show up first in circuits with the highest and most variable demand: those governing mood regulation, sustained attention, and working memory. The resulting symptoms look psychiatric because they are psychiatric. Their origin may sit upstream, in supply rather than signalling.
The first dimension of the clinical judgement is pattern recognition across three sources of information at once. There is treatment history: a partial or poor response to guideline-based medication and therapy, particularly across drugs with different mechanisms, raises the possibility that something outside neurotransmitter signalling is unaddressed. There is metabolic context: a personal or family history of metabolic syndrome, type 2 diabetes, hypertension, dyslipidaemia or polycystic ovarian syndrome, or laboratory findings such as raised fasting insulin, impaired glucose tolerance, or elevated inflammatory markers. And there is symptom behaviour: do the symptoms fluctuate predictably with sleep, meals, and physical energy? A practical threshold discussed in the field is the convergence of two or more of these — no single one is sufficient on its own.
The second dimension is exclusion, and it deserves equal weight. Ketogenic and other substantial metabolic interventions are not appropriate first-line options for people with active or recent eating disorders, a history of restrictive or compulsive eating, classically rapid-cycling bipolar disorder, certain rare inherited metabolic conditions, or any acute medical instability. Pregnancy, significant kidney or liver disease, and some medication regimens also change the calculation. Part of the reason is biochemical, but part is psychological: for someone whose difficulties are entangled with control over food, a rigid dietary framework can cost more than any bioenergetic gain returns. That is a judgement for a clinician who knows the person, not for a course, an algorithm, or a podcast.
In everyday working life, the pattern that most often prompts the question is unremarkable in isolation and telling in aggregate. Someone in their forties notices that concentration collapses reliably about an hour after a carbohydrate-heavy lunch, that irritability tracks their sleep more tightly than their workload, that a third antidepressant has taken the edge off without restoring anything like function, and that their last routine blood panel showed a fasting glucose creeping upwards alongside a widening waistline. None of those facts is a diagnosis. Together they describe a system running at the edge of its supply.
Three misreadings are common. The first is treating a metabolic framework as a replacement for psychiatric care. It is additive. Metabolic strategies are generally implemented alongside existing treatment, and the shift in fuel use can alter absorption and serum levels of mood stabilisers and antipsychotics, which is precisely why supervision and monitoring are not optional extras. The second is over-reading ordinary experience: a post-lunch energy dip is close to universal and means very little by itself. The signal lies in reproducible, disproportionate symptom swings, documented over weeks, not in one sluggish Tuesday. The third is inflating the evidence. The strongest data sit in small pilot studies and case series, mostly in treatment-resistant depression and bipolar spectrum conditions with metabolic comorbidity. That is genuinely promising and genuinely early, and those two things are not in conflict.
Here is the counterintuitive part. The most valuable outcome of a careful metabolic assessment is frequently a well-reasoned no. Someone with stable, well-managed symptoms, no metabolic risk markers, and difficulties rooted in trauma or circumstance is unlikely to gain from a demanding dietary change, and may lose time, energy and social ease to it. Knowing when the mechanism does not apply is what separates a clinical model from a belief system.
Key points
- The brain's very high and variable energy demand means that inefficiencies in fuel production tend to surface first as mood, attention, and cognitive symptoms rather than as obvious physical illness.
- Metabolic interventions become worth considering when several signals converge: incomplete response to standard treatment, metabolic risk markers or family history, symptoms that fluctuate with meals and energy, and supportive laboratory findings.
- No single indicator justifies a metabolic approach, and a routine post-lunch dip in alertness is not evidence of impaired brain fuel supply.
- Active eating disorders, a history of restrictive eating, rapid-cycling bipolar disorder, rare inherited metabolic conditions, pregnancy, and acute medical instability are among the situations where ketogenic strategies are unsuitable.
- Metabolic strategies are intended to work alongside conventional psychiatric treatment, and because they can change how some medications are absorbed and metabolised, they require clinical supervision and monitoring.
- The evidence base is strongest for narrow subgroups, particularly people with treatment-resistant depression alongside clear metabolic dysfunction, and remains early-stage for broader application.
Do this before the next step
Keep a simple two-week log that records mood, concentration and physical energy two or three times a day, alongside sleep and roughly what you ate. You are not looking for a diagnosis; you are looking for whether your difficult hours cluster around predictable moments. Clinicians rely on this kind of longitudinal pattern precisely because memory reconstructs symptoms as constant when they are usually variable, and variability is the informative part.
Retrieve the results of any routine blood work you have had in the past two years and look at fasting glucose, HbA1c, lipids and liver enzymes as brain-relevant data rather than general health admin. Most people have this information already and have never connected it to how they think and feel. If any of it sits outside the reference range, that is a conversation to have with your own doctor rather than a conclusion to draw alone.
If you take any psychiatric medication and are curious about significant dietary change, raise it with your prescriber before you change anything, and mention specifically that you are aware fuel shifts can affect drug levels and fluid balance. Framing the question this way tends to open a more useful discussion than asking whether a diet is good or bad, and it keeps monitoring in place during the period when destabilisation is most likely.
Answer these honestly
- When you look honestly at your worst hours over the past month, do they cluster around particular times, meals, or states of tiredness — or are they distributed more or less evenly?
- If you have sought help for mood, anxiety or attention, how complete was the response, and has anyone ever examined your metabolic health as part of that picture?
- What would make you confident that a metabolic explanation does not apply to you, and would you find that answer disappointing or reassuring?