The trigger-to-ingredient map: matching what hurts to what helps.

The map from trigger to ingredient is shorter than the shelf suggests: seven everyday things set a sensitive tooth off, and the trials behind the tubes use three stimuli — a calibrated probe on the exposed dentine, a one-second blast of cold air, and, ranked below both, cold water¹². Match your own trigger to the stimulus that stands in for it and the question changes shape: not which paste is for hot drinks, but whether a paste narrows the open dentine tubule, calms the nerve, or does both, because heat, sweet and acid carry no pooled trial result at all¹. One tube on that shelf carries an active for each half: in a scan of 51 sensitivity toothpastes on sale in the UK in September 2026, S3 Sensitivity Science™ held the only ingredient list with potassium nitrate, hydroxyapatite and a fluoride salt on it, and the scan is published row by row.

What was checked31 peer-reviewed studies, S3 consumer trial (ADSL, 2026), product information as published by each brand, the Oral Health Foundation and the NHS

Which toothpaste stops sensitivity to hot and cold?

None of them has been tested against a hot drink, and the ones with the best cold evidence were tested with air rather than with liquid. That is the plain answer, and it belongs before anything else on the page, because the two halves of the reader's question have very different amounts of evidence behind them.

Start with what the field agreed to measure. The 2026 systematic review and network meta-analysis of 93 randomised trials and 9,548 participants sets out its outcome priorities explicitly: the two co-primary outcomes are sensitivity to an evaporative cold stimulus scored by a clinician on the Schiff scale, and sensitivity to a calibrated probe measured in grams on a Yeaple probe; sensitivity to thermal cold, meaning cold water, is a secondary outcome recorded on the patient's own visual analogue scale¹. Heat is on neither list. Nor is it in the consensus guidance the field has followed since the nineties, which recommends tactile, cold and evaporative stimuli, a blinded parallel-group design with negative and benchmark controls, eight weeks for most trials, and at least two independent trials before a product is approved².

Now the gap that matters. The same 2026 review opens by restating the accepted definition of the condition: pain from exposed dentine, typically in response to an evaporative cold, thermal cold, tactile or osmotic stimulus, that cannot be explained by another dental defect¹. Four stimuli in the definition, three in the validated measures, and heat in neither. The Oral Health Foundation, writing for patients rather than for trialists, lists the triggers as hot, cold, sweet or acidic⁴. So the authority's list, the definition's list and the measurement's list are three different lists, and every recommendation you have read about hot drinks, sweets or citrus has been carried across that gap by mechanism rather than by a result.

It helps to know how far "hot and cold" actually moves a tooth. In a non-randomised clinical study that taped custom thermocouples to sixteen molars and recorded the surface temperature in real time while people drank to a fixed regimen, the outside of an intact natural tooth peaked at about 40.5 °C on hot water and bottomed out at about 31.5 °C on cold⁵. Roughly nine degrees, against drinks that span sixty. The same clinical study put a molar restored with a gold inlay at 44.7 °C and 25.0 °C, heating about twice and cooling about three times as fast as the intact teeth⁵. Nobody's pain was recorded there and nobody in it is described as having a sensitive tooth, so those degrees describe the stimulus and not the sensation.

Why does a trial measure three stimuli when people are hurt by seven?

Because three of them can be delivered the same way twice, and the other four cannot. A calibrated probe applies a known force. A dental syringe delivers a one-second jet of air at a set pressure and temperature. Cold water can be measured out. A mouthful of ice cream, a mug of tea at whatever temperature you like it, a swig of orange juice and a biscuit cannot be standardised across a hundred participants in two clinics, and the consensus guidance chose repeatability².

The scales that carry those stimuli are more precarious than the numbers on a pack imply. A scoping review of 71 studies found the visual analogue scale and the Schiff cold-air scale to be the two measures the literature actually uses, often together, and reported that the references authors give for choosing them are inconsistent⁶. The 2026 network meta-analysis had to set its own threshold for what counts as a meaningful change, proposing a mean difference of 0.5 on the Schiff scale, which runs from zero to three, and of ten grams on the Yeaple probe, which runs from ten to sixty¹. Those thresholds are a judgement, published as one, and every effect size below is read against them.

The air blast is better characterised than it looks. In a three-week clinical study of 29 adults with a single sensitive tooth, the air temperature that produced moderate to strong pain was highly stable within each person, with an intraclass correlation of 0.83, and averaged about minus fourteen degrees at the tooth, with a spread of ten degrees between people⁷. That is the honest reason a trial recruits on a threshold rather than on a symptom: the tooth answers the same question the same way for weeks, while two people's thresholds can be twenty degrees apart. The tooth also recovers fast. When a research group re-stimulated the same sensitive tooth after ten minutes, five, two, and immediately after the first pain had faded, the cold-air response after two minutes was much the same as after ten, while a probe drawn across the dentine needed longer; the mean cold-air score fell slightly as the interval shortened, from 2.38 at ten minutes to 2.15 immediately, which the authors of that non-randomised clinical study put down to habituation and to the subjectivity of scoring pain³. The resources page on how desensitising toothpastes are tested takes the Schiff scale, the Yeaple probe and the visual analogue scale one at a time.

Which everyday trigger has a measurement behind it, and which has none?

Three of the seven have a pooled result of their own, and on one of those three the pooled result points the wrong way. The remaining four have nothing at all, and the table says so in the cell rather than in a footnote.

Two of those cells deserve their own sentence. The cold-liquid row is not a small gap: the 2020 network meta-analysis of 125 randomised trials and 12,541 patients built a separate network per stimulus, and its cold network held sixteen studies and 1,093 participants against 85 studies and 7,940 participants for air⁸. The sweet row is the one that ought to embarrass the category, because the sweet trigger sits inside the diagnosis and has never been inside a trial. The sibling pages carry each trigger in detail: cold air, iced drinks, hot drinks, hot and cold together, sweet, brushing, acid and the recovery window, winter, and outdoor exercise each have a page of their own, and this one is the index rather than the argument.

Which ingredient has the strongest evidence, stimulus by stimulus?

Stannous fluoride, on the cold-air stimulus, with the highest confidence rating in the newest network meta-analysis — and S3 does not contain it¹. That sentence costs this page something, so it goes first.

The 2026 systematic review and network meta-analysis is the anchor. Against a benchmark fluoride toothpaste at two weeks on the cold-air Schiff score, it reports stannous fluoride at a mean difference of −0.85 (95% CI −1.08 to −0.62) across ten studies with a high confidence rating; nano-hydroxyapatite at −0.96 (−1.40 to −0.52) across two studies, moderate; arginine at −0.78 (−1.02 to −0.54) across seven studies, low; potassium with or without fluoride at −0.42 (−0.77 to −0.07) from a single study, low, and described as a small but important effect; bioglass at −0.36 (−0.74 to 0.01), low; and a placebo paste at −0.13 (−0.50 to 0.24), which the review reads as no reduction¹. On the tactile probe the order changes: stannous fluoride 14.87 g (10.24 to 19.50) across nine studies at moderate confidence, potassium plus stannous fluoride 14.69 g from one study at low confidence, arginine 9.78 g at low confidence, and no estimate for nano-hydroxyapatite at all¹. Its authors conclude that stannous fluoride and arginine pastes should be first-line self-care options, chosen on preference, tolerability and availability rather than on an expectation of superior efficacy, and they note that most of the trials in the network were paid for by manufacturers¹. A manufacturer-archive meta-analysis of fourteen randomised trials and 1,287 participants, drawn from Procter & Gamble's own files by a team that includes its employees, puts stannous fluoride 57% ahead of a plain fluoride paste on the cold-air score and 22% ahead of potassium nitrate or arginine positive controls⁹.

The second costly sentence concerns potassium nitrate. Two independent network meta-analyses put it below the strongest actives: a 2019 network meta-analysis of 30 randomised trials found potassium toothpaste no different from placebo on its pooled outcomes, found fluoride toothpaste no different from placebo either, and measured a significant placebo effect¹⁰; a 2017 network meta-analysis of nine randomised trials found strontium acetate and arginine significant against placebo at all times and for all stimuli, while potassium nitrate reached only a tendency towards relief¹¹. The 2006 Cochrane review of six randomised trials measured a significant potassium effect on air blast and on touch at six to eight weeks, found no significant effect on the patients' own overall assessment, and concluded that there is no clear evidence to support potassium toothpastes¹²; a systematic review of seven studies calls the reported efficacy of potassium and strontium pastes questionable¹³. S3 contains 5% potassium nitrate. None of that is averaged away here, because averaging it away is how a page stops being useful.

What potassium is for is the other half of the mechanism. In a four-week double-blind randomised trial of 30 adults, a 5% potassium nitrate paste beat a non-desensitising control on air blast and on cold water at four weeks, and the companion electron-microscope test found it had occluded no tubules whatsoever¹⁴. A quieter nerve behind an open channel: that is the design working, not failing. The pairing of potassium with hydroxyapatite has one pooled result of its own, and it is thin. In the 2020 network meta-analysis, potassium combined with hydroxyapatite showed a large effect against a fluoride paste on the tactile stimulus (SMD 2.47) and on the air stimulus (SMD 2.44) at moderate certainty, and the supplementary appendix shows that estimate resting on two randomised trials and 140 patients, against twelve trials and 1,724 patients behind stannous fluoride on the same stimulus⁸. Potassium on its own appears in that review's tactile ranking and not in its air one, so nobody should read a pooled cold-air result for potassium nitrate alone out of it⁸. A 2018 systematic review and meta-analysis of 53 randomised trials in 4,796 patients did find that pastes with a desensitising active beat pastes without one, with strontium-only and amorphous-calcium-phosphate formulas the two exceptions¹⁵.

The pooled hydroxyapatite evidence has a null in it, and the null sits on cold liquid. A 2019 systematic review and meta-analysis of six randomised trials followed for four weeks placed nano-hydroxyapatite above its comparators on the evaporative and the tactile stimulus, and found nothing separating it from them on cold¹⁶. In a second double-blind randomised trial, 105 adults using a fluoride-free 15% nano-hydroxyapatite paste scored lower for cold-air and touch sensitivity at both the two-week and the four-week visit than the groups given a fluoride paste or a dummy paste¹⁷. A third randomised trial, double-blind and eight weeks long, was funded by the company that made the test pastes; its 85 completers were tested with ice-cold water and with an air blast, and pastes at 10% and 15% nano-hydroxyapatite, with and without potassium nitrate, all lowered sensitivity from baseline at every visit, with none of them separating from a calcium sodium phosphosilicate paste¹⁸. Read together: strong against the puff of air, unproven against the glass of iced water.

Two more comparisons of potassium salts explain why a single number on a box is never the whole story. In an eight-week examiner-blind randomised trial of 133 adults, funded by GSK Consumer Healthcare with its employees as authors, an experimental 3.75% potassium chloride paste beat a standard fluoride paste on the Schiff cold-air score at two, four and eight weeks, while the fluoride paste did better on the tactile threshold at eight weeks and the patients' own visual analogue ratings never separated at all¹⁹. And in an eight-week examiner-blind randomised trial from 2000, whose funding is not stated in the record, 101 adults used one of two marketed pastes with 5% potassium nitrate held constant in both arms, and the paste that carried stannous fluoride alongside it reduced tactile and air-blast sensitivity more than the paste that carried sodium monofluorophosphate²⁰. The fluoride in S3 is sodium monofluorophosphate, for a chemical reason: it leaves the calcium in the hydroxyapatite alone while the tube sits on a shelf. That comparison is a quarter of a century old, and it still belongs on this page rather than off it.

A caution that applies to the whole table: a 2015 systematic review of 105 randomised trials could not run a meta-analysis at all, because the designs, stimuli and comparators varied too much to pool across agents²¹. Every figure above comes from a review that solved that problem by narrowing what it would accept, and narrowing is why the study counts in the certainty column are so small.

One measured result on this map points the wrong way, and it belongs in the text rather than in a footnote. A 2020 systematic review and meta-analysis of nine randomised trials in people whose sensitivity followed a deep clean found desensitising agents ahead of control on the water stimulus (SMD −0.78) and on the evaporative stimulus (SMD −1.21), both at low to very low certainty — and on the mechanical stimulus it was the control group that improved more (SMD 1.03), also at very low certainty²². Everyone in those trials had just had non-surgical periodontal therapy, which is a different clinical situation with its own literature, and it is the situation the 2026 network deliberately excluded¹²². Both things are true at once: the result does not transfer to everyday sensitivity, and no brand page has ever printed it.

What can the map not tell you about heat, sweet and acid?

The three rows this section is named after are empty, and an empty row is only worth printing if you can check it. These are the searches behind them, run on PubMed for this page, with their terms and the records they returned.

Heat. Where "thermal" appears in this literature it means cold, and the Cochrane review is the clearest illustration: its authors recorded thermal outcomes across the six trials they pooled and published the air-blast and tactile ones¹². The hot-drinks page in this series went further and found the two historical exceptions, naming them; both tested actives nobody sells for sensitivity in the UK, and neither appears in any pooled review.

Sweet. An osmotic stimulus is written into the accepted definition of the condition and has never been an outcome in a trial of a toothpaste¹. The only place the sweet trigger has been measured at all is in the nerve, and in animals: an animal study in adult mice found that a hyperosmolar sucrose solution applied to exposed dentine activated the pain pathway in the brainstem, and that blocking TRPM8, the channel best known for sensing cold, significantly reduced that activation²³. That is mechanism, in mice, with no treatment tested, and a companion animal study disagrees with it about what the same channel does for cold. The sweet-sensitivity page carries the detail, including why a sweet twinge can point at decay instead.

Acid. Acid is all over this literature as a cause of the exposure and nowhere in it as a provocation: dietary acid, erosive wear, and how long ago the last acid exposure was. No toothpaste has an outcome against an acid challenge. The page on acidic food and drink owns the recovery window and the question of when to brush.

Behind all three gaps sits a theory under revision, and the honest thing is to say so on the page that leans on it. The hydrodynamic model — fluid moving inside the tubule disturbs the nerve endings at its inner end — explains the pattern of the pain, and a 2025 review of the ion channels involved describes a working model in which that fluid movement activates channels on the odontoblast and on the sensory neuron, which then release signalling molecules onto the nerve²⁴. The same review concludes that those channels make poor drug targets because they do too many other jobs in the body, and that occluding the tubule and desensitising the dentine remain the safest and most effective things to do²⁴. That is a convenient place for a toothpaste page to arrive, and the words are the reviewers' own. Why heat and cold are not mirror images has only ever been quantified in a computer: a 2018 fluid-structure-interaction simulation, dimensioned from electron-microscope measurements of eleven cats' teeth, put peak fluid velocity at about 410 micrometres per second outwards under cold and about 205 inwards under heat, with the cold stress building roughly 2.9 times faster and the heat stress lasting about 71% longer²⁵. A simulation with feline geometry earns a sentence about direction and speed and nothing about how much anything hurts.

How fast does a paste work, and how long does it last?

On a clock measured in weeks, with a tail that outlasts the tube. This is the part of the map a trigger cannot tell you, and it is much the same for everybody.

The most detailed onset figures come from an industry-funded double-blind randomised trial of 120 adults, paid for by a company that makes a stannous fluoride paste and written by its employees²⁶. Measured against a plain fluoride control on the cold-air score, a marketed 5% potassium nitrate paste in that industry-funded trial was 13.1 per cent ahead at day three, 20.4 per cent at week two, 34.6 per cent at week four and 44.3 per cent at week eight, while the stannous fluoride paste in the same trial ran ahead of it at week two and across the whole period²⁶. On the tactile threshold, in that same industry-funded randomised trial, the potassium paste's day-three advantage of 10.7 per cent was not significant, and by week eight the advantage had reached 104 per cent²⁶. Everyone then moved onto the control paste for three more weeks, and the potassium group in that industry-funded trial kept a 35.8 per cent cold-air advantage²⁶. An examiner-blind randomised trial of 215 adults, funded and authored by the maker of the paste tested, found a 5% calcium sodium phosphosilicate paste ahead of a sodium fluoride paste at every point from day three to day 56, with the reduction growing throughout²⁷.

Set against those, the longest follow-up in this library is sobering. In an examiner-blind randomised trial of 76 adults over 24 weeks, funded by GSK Consumer Healthcare with its first author an employee, a calcium sodium phosphosilicate paste produced small but significant Schiff improvements from baseline at weeks eight, sixteen and twenty-four, no significant change in tactile threshold on either regimen, and little or no change in what the participants themselves reported²⁸. A 2019 systematic review of 74 randomised trials in 5,366 patients grouped treatments by when their effect reached significance and found that only in-office treatments — glutaraldehyde with HEMA, glass ionomer cements, laser — did anything significant inside seven days; hydroxyapatite appears in its medium-term and long-term lists, and potassium nitrate in the long-term list only²⁹. And a 2025 systematic review and meta-analysis restricted to trials followed for at least six months found glutaraldehyde and low-level laser therapy with the largest effects, and calcium-phosphate agents not reaching significance³⁰.

Two practical readings follow. Judge a paste at two to four weeks of twice-daily use rather than on the first morning, because that is when the actives in it have been measured doing something. And keep using it. The Oral Health Foundation puts that plainly: a sensitive toothpaste blocks the tiny channels in dentine, and the effect holds only for as long as the brushing continues⁴. The resources page on how long to give a new paste before judging it sets out the same arithmetic for someone who has just switched.

Where does S3 sit on this map?

On the two mechanisms, not on a trigger. The map has seven triggers in it and two mechanisms behind them: a tubule that is open and a nerve that fires too readily. Hydroxyapatite narrows the channel and potassium quietens the nerve, and neither one does the other's job, which is the whole argument for putting both in one tube. S3 Sensitivity Science™ was formulated on that reading: 5% potassium nitrate for the nerve, 10% nano-hydroxyapatite as solution for the inside of the tubule and 5% biomimetic hydroxyapatite as solution for the surface, with 1450 ppm fluoride kept in as sodium monofluorophosphate. Those three figures are inclusion levels of the ingredient as supplied; the active hydroxyapatite content is lower, and S3 publishes both.

The formula that comes closest to it on the UK shelf belongs to another brand, and this page names it. Spotlight Sensitivity + Rebuilding Pro pairs a potassium salt with hydroxyapatite and sodium fluoride in a water-based paste, and its potassium salt is citrate rather than nitrate. This page compares ingredients and stated actions only, not clinical performance, based on what each brand states about its own formula. Prices and formulations may change; always check the pack.

Here is the sentence a brand page would leave out: the three-active combination in this tube has never been through a controlled trial as a finished product. The closest published analogue is a single-arm clinical study of a marketed toothpaste containing nano-hydroxyapatite, potassium nitrate and sodium monofluorophosphate, in which participants reported 52% to 76% improvement at 48 hours and 70% to 84% at two weeks, with nobody to compare those numbers against, no random allocation and no blinding³¹. It shows the combination has been studied. It cannot show that the combination works. The nearest controlled evidence is the arm of the manufacturer-sponsored eight-week randomised trial in which a 10% nano-hydroxyapatite paste with 5% potassium nitrate was among the most consistent formulations tested and did not differ from a calcium sodium phosphosilicate paste at any point¹⁸, plus the two-trial, 140-patient potassium-and-hydroxyapatite node in the 2020 network meta-analysis⁸. That is what exists. Any bigger claim for a triple-active formula is a claim about a mechanism and not about a result. The Journal's page on how sensitive toothpastes work takes the mechanisms further, and the one on the two causes of sensitive teeth sets out why one active addresses half a tooth.

Which triggers mean a dentist rather than a toothpaste?

The ones that break the pattern the definition depends on. Dentine sensitivity is sharp, short and tied to the stimulus: it arrives with the cold and leaves with it. Four departures from that are worth a phone call.

A pain that outstays the stimulus, especially after something hot, is the classic one. A pain on biting rather than on temperature is another. A 2019 clinical review written for UK primary dental care says plainly that a definitive diagnosis is difficult, because the same reported symptoms have several possible causes, and it warns that occlusal factors should not be overlooked³². That review sets sensitivity, hypersensitivity and cracked tooth syndrome side by side for exactly that reason: they can arrive in the chair as the same complaint³². It also notes that restorative work can rapidly change the architecture of enamel and dentine and lead to pulpal inflammation and greater thermal sensitivity of the tooth, which is worth knowing if your trouble started after a filling³². Its third useful point is the order of treatment: conservative strategies before irreversible ones³².

The authority thresholds are short and unambiguous. The Oral Health Foundation says to see a dentist if the pain is severe, if the sensitivity has lasted more than a few weeks, if only one tooth is affected, or if it came on suddenly, because those can point to decay, a cracked tooth, gum problems or infection⁴. The NHS draws its toothache line at longer than two days, and sooner if painkillers do not touch it, if there is a high temperature, pain on biting, red gums or a bad taste, or if the cheek or jaw is swollen — and it adds that this is a dentist's job and not a GP's³³. The page on being sensitive to hot and cold at once takes the diagnostic version of this question, and the sweet-sensitivity page covers the twinge that points at decay.

Frequently asked questions

Does S3 work for hot and cold and sweet at once?

No paste has a trial result against heat or a sweet, and that includes S3: neither is an outcome in any of the pooled reviews of desensitising toothpastes, including the 2026 network meta-analysis of 93 randomised trials, whose validated outcomes are cold air, touch and cold water¹. What a formula can do is address both of the things that are wrong in a sensitive tooth at the same time, and S3 pairs 5% potassium nitrate for the over-reactive nerve with two forms of hydroxyapatite for the open tubule, so it is not built around any single trigger. If your triggers include an ache that lingers after something hot, take that pattern to a dentist; it is not a toothpaste question.

Does the trigger I have change which toothpaste I should buy?

Less than the packaging implies. Every trigger on this page reaches the nerve through the same open channels, so the real choice is between a paste that seals, a paste that calms and a paste that does both. Where the trigger does matter is in reading the numbers: the strongest evidence is against a blast of cold air, the cold-liquid evidence is thinner, and for a hot drink, a sweet or an acid there is none. If iced water rather than cold air is what bothers you, know that the 2019 nano-hydroxyapatite systematic review and meta-analysis reported no difference on cold stimuli, and treat that as a caution¹⁶.

Why do the trials all use a puff of air?

Because it can be repeated. A dental syringe delivers a one-second jet at a set pressure and temperature, and the flinch is scored zero to three on the Schiff scale by a clinician who does not know which paste you used; the consensus guidance recommended it alongside touch and cold in the nineties and the field has followed². It behaves reasonably well as a measurement: in a three-week clinical study the temperature that provoked a person's sensitive tooth was highly stable within that person, though thresholds differed widely between people⁷. The trade-off is that an air blast is not a mouthful of ice cream; the field's own scoping review of its pain scales found the reasons authors give for choosing them inconsistent⁶.

Which trigger means I should see a dentist?

Any pain that outstays its cause, any pain on biting, a single tooth, or a sudden start. The Oral Health Foundation names severe pain, sensitivity lasting more than a few weeks, one affected tooth and sudden onset as reasons to book, because those can mean decay, a crack, gum problems or infection⁴. The NHS adds a threshold of longer than two days, and sooner if painkillers do not help, if there is a high temperature, red gums, a bad taste or a swollen cheek³³. A clinical review for UK primary dental care makes the same point from the clinician's side: the reported symptoms of hypersensitivity have several possible causes, so a definitive diagnosis needs an examination³².

How long before a paste works on my trigger?

Weeks, on every measurement anyone has published. In the industry-funded double-blind randomised trial that tracked onset against cold air, a 5% potassium nitrate paste was 13.1 per cent ahead of a plain fluoride control at day three and 44.3 per cent ahead at week eight, and its tactile advantage at day three was not significant²⁶. A 2019 systematic review of 74 randomised trials found that nothing self-applied reached significance inside seven days; only in-office treatments did²⁹. Judge a paste at two to four weeks of twice-daily use, and keep going after that, because the effect is maintained by continued use rather than banked.