Overheating in homes, and what measurement can't tell you

The heatwave of 9–15 August has ended. It peaked at 38.1°C at Kew Gardens on the 13th — the fifth-hottest reading recorded in the UK, and well short of the 40.3°C recorded at Coningsby in July 2022. Amber heat-health alerts covered all nine English regions from 10 August, were downgraded on the 15th, and cleared.

Every summer this leaves housing teams with the same question: which of our homes overheat? This page is about why that question is much harder to answer than it looks, and why the measurement Make Sense is building does not answer it.

First, the mortality figures, because they are routinely misread

UKHSA's heat mortality reports estimate around 2,985 excess deaths in England in summer 2022, 2,295 in 2023, 1,311 in 2024 and 1,504 in 2025.

These are not counted deaths. No cause-of-death coding is involved. A heat episode is defined as any day where mean Central England Temperature reaches 20°C, and the estimate is the gap between observed daily deaths and a rolling baseline. UKHSA states plainly that heat-associated deaths "cannot be measured directly but only estimated" (methodology).

How much does the method matter? For the same summer — 2022 — three published approaches give 2,985 (UKHSA, England), 3,271 (ONS/UKHSA, England and Wales) and 4,507 (ONS climate-related mortality, a different model). Rapid attribution studies are a fourth method again, with much wider uncertainty.

So: quote the figure with the method attached, and use "estimated excess deaths during heat periods" rather than "deaths caused by heat". That distinction gets lost frequently, including in official press releases.

Two things worth carrying into policy. Most of these deaths occur in care homes and hospitals, not outdoors. And in 2024, significant mortality occurred under yellow alerts — the alert levels are impact-based judgements, not temperature triggers, and moderate heat is not safe heat.

Part O does not apply to your existing stock

This is the single most consequential fact on this page.

Approved Document O — overheating — was made by SI 2021/1391 and came into force on 15 June 2022. It requires new residential buildings to limit solar gains and provide a means of removing excess heat, demonstrated either by a simplified method or by dynamic thermal modelling to CIBSE TM59.

Government guidance is unambiguous that "Part O only applies to new residential buildings". It does not apply to extensions or conservatories. It is absent from Regulation 6 of the Building Regulations 2010, which means it does not bite on a change of use — so an office-to-residential permitted development conversion, arguably the highest-risk housing type in the country for overheating, escapes it entirely (Part O FAQ).

Both the Environmental Audit Committee and the Climate Change Committee have recommended extending it to refurbishment. That has not happened.

The practical consequence: the regulation addresses the homes being built, and the overwhelming majority of the risk sits in the homes already standing. There is no equivalent duty, and — per the CCC's 2025 progress report — no national dataset tracking it. Policy is rated "limited"; delivery is rated "unable to evaluate".

Nobody agrees how many homes overheat, and the numbers are not comparable

The English Housing Survey 2024–25 reports around 3 million homes, 12%, as uncomfortably hot in summer, up from 7% in 2019. But that is self-reported perception, and it inverts the modelling consensus: detached houses (15%) and bungalows (14%) reported most, purpose-built flats (9–10%) least. Modelling consistently finds top-floor flats among the highest risk.

Measured data does not settle it either. Analysis of the Energy Follow-Up Survey found 2.5% of homes failed CIBSE TM59 Criterion 1 while 26% failed Criterion 2 — a tenfold difference from the same dataset depending on which threshold you apply.

The CCC's widely quoted "92% of homes could overheat" is a 2050 scenario under hotter heatwaves, not a present-day count.

Treat every headline figure here as an artefact of its metric. If you are commissioning work, the threshold definition matters more than the sample size.

Does insulation cause overheating? Mostly, no — and it is a minor term

This is contested in public and much less contested in the literature.

The largest controlled study — Fosas et al., Building and Environment (2018) — isolated insulation across roughly 576,000 building variants spanning climate, thermal mass, glazing, shading and ventilation. It found insulation "only accounts for up to 5% of overall overheating response", and that in cases not already overheating through poor design there is a strong tendency for increased insulation to reduce overheating.

The physics is symmetrical: insulation slows heat flow in both directions. In a UK heatwave, outdoor air exceeds indoor for only a few hours a day, so fabric conduction is a modest inbound path. Insulation cuts daytime ingress but also slows the night-time discharge of solar and internal gains. Which effect dominates depends on the rest of the building.

Ranked by influence, the drivers are roughly: solar gain (glazing ratio, orientation, shading), then ventilation capacity — including whether windows can actually be opened given noise, security and safety — then internal gains, of which uninsulated communal heating risers in flats are a significant and under-recognised source, then thermal mass, and only then insulation.

There is genuine contrary evidence: measured work on UK Passivhaus dwellings has found a majority of bedrooms failing TM59 night-time criteria even in mild summers, and some retrofit monitoring indicates added insulation raises warm-season indoor temperatures. The reconciliation most consistent with the evidence is that harm comes from the package rather than the insulation — airtightness without commissioned purge ventilation, mechanical ventilation with no summer bypass or with the bypass unused, loft and room-in-roof conversions, single-aspect layouts.

And in typical years, modelling indicates heat-mortality increases from efficiency retrofit are more than offset by reductions in cold mortality. Cold still kills more people in this country than heat.

All of which is to say: "insulation causes overheating" is not supported, "insulation prevents overheating" is not supported either, and anyone offering you a confident one-line answer is not reading the same evidence.

What our kind of measurement will not tell you

We should be direct, because this is where a supplier would normally start selling.

Make Sense is building toward street-level thermal measurement of building heat loss. That has no validated role in assessing overheating risk.

External infrared thermography is a winter diagnostic. The governing standard, BS EN ISO 6781-1:2023 — which superseded BS EN 13187 in August 2023 — frames it as detection of thermal irregularities, and is explicit that it does not determine the degree of insulation. It requires a sustained indoor-to-outdoor temperature difference and no direct solar radiation on the surface. Summer surveys are unreliable for exactly that reason: in July the surface signal is dominated by incident sunlight.

More fundamentally, it measures the exterior surface temperature of a wall. Overheating, as defined by TM52, TM59 and Part O, is about operative temperature inside occupied rooms during occupied hours. A hot façade in August tells you about solar load on that elevation. It tells you nothing about whether a bedroom exceeded 26°C at 3am.

There is one narrow, honest inferential use: thermography can reveal uninsulated communal heating risers in blocks of flats, which are a real internal-gains problem. That is a fabric-defect finding that happens to be relevant to overheating. It is not an overheating assessment.

What actually answers the question is indoor temperature logging in occupied rooms over a summer, assessed against a stated TM59 criterion. That is a different instrument, a different deployment and a different supplier.

Where this leaves an authority

If overheating is on your risk register, the useful next steps are unglamorous: decide which criterion you are assessing against before you commission anything; look hardest at top-floor flats, single-aspect layouts and change-of-use conversions, none of which Part O reaches; and treat any survey that does not measure indoor temperature over time with scepticism, including ours.

We write about the gap between modelled and measured building performance in the housing energy performance gap. Overheating is the same problem wearing different clothes: a modelled answer standing in for an observation nobody has made.

Make Sense holds no certification, has no deployed system and publishes no performance figures. On this particular question we are not the right supplier, and we would rather say so here than in a procurement. If you want to talk it through anyway, get in touch.

---

Sources

• UKHSA — heat mortality monitoring reports
• UKHSA — heat mortality monitoring: quality and methodology information
• UKHSA — heat-health alerting system guidance
• The Building Regulations etc. (Amendment) (England) Regulations 2021, SI 2021/1391
• Approved Document O — overheating
• Approved Document O — frequently asked questions
• Climate Change Committee — A Well-Adapted UK (CCRA4), May 2026
• English Housing Survey 2024–25: weather resilience in housing fact sheet
• Fosas, D. et al. (2018) "Mitigation versus adaptation: does insulating dwellings increase overheating risk?", Building and Environment 143, 740–759
• Petrou, G. et al. (2019) analysis of overheating in the Energy Follow-Up Survey
• BS EN ISO 6781-1:2023 — detection of heat, air and moisture irregularities in buildings by infrared methods
• Met Office — what is a heatwave