Very hot drinks linked to more than three times the risk of squamous cell oesophageal cancer: study of nearly a million Britons

A preference for drinking tea, coffee, and other hot beverages at very high temperatures has been linked to a significantly increased risk of squamous cell carcinoma of the esophagus. In a new prospective study of nearly 980,000 UK residents, those who described their usual beverage temperature as "very hot" had a 3.17 times higher risk of this type of cancer, after accounting for the number of cups consumed, than those who preferred warm drinks. The study was published on September 8, 2026, in the International Journal of Cancer.

The association showed a clear dose-response relationship. Even the "hot" category was associated with a roughly 69% higher risk compared to "warm" drinks, and for "very hot" drinks, the relative risk was 3.17. Moreover, the risk of developing the disease increased not only with the temperature but also with the number of cups consumed daily. Six or more hot drinks per day were associated with a 71% higher risk of squamous cell carcinoma, even after accounting for preferred temperature.

Moreover, the researchers discovered a fundamentally important difference between the two main histological types of esophageal cancer. Temperature had virtually no association with esophageal adenocarcinoma: for "very hot" drinks, the relative risk was 0.92 with a 95% confidence interval of 0.76-1.12. This contrast supports the hypothesis that the primary mechanism is related to repeated thermal injury to the squamous epithelium of the esophagus, rather than the chemical components of tea or coffee.

The authors calculated that, if the discovered association is truly causal, approximately 14% of cases of squamous cell esophageal cancer in the UK could theoretically be prevented if people who prefer their drinks "very hot" allowed them to cool to at least the "hot" category. Converted to national incidence, this equates to approximately 440 cases per year. However, the study remains observational and does not prove causality with the same certainty as a randomized trial.

Key indicator	Result
Total participants	977 282
Million Women Study	510 112
UK Biobank	467 170
Average observation	14.2 and 11.1 years
Total cases of esophageal cancer	2348
Squamous cell carcinoma	1045
Adenocarcinoma	1303
"Hot" vs. "Warm," squamous cell carcinoma	OR 1.69
"Very hot" versus "warm"	OR 3.17
6+ drinks/day vs. <6	OR 1.71
"Very hot", adenocarcinoma	OR 0.92
Potentially preventable proportion of cases	≈14%
Estimating cases in the UK	≈440 per year
DOI	10.1002/ijc.70654

Why were the researchers interested in temperature, and not tea or coffee?

The link between very hot drinks and esophageal cancer has been studied for several decades. The most compelling observations previously came from Iran, South America, China, Japan, and several African countries, where tea, mate, and other beverages are often consumed at approximately 70°C. In 2016, the International Agency for Research on Cancer classified the consumption of beverages above approximately 65°C as "probably carcinogenic to humans."

But this classification left an important question for Europe and North America. Tea and coffee are also very frequently consumed there, but the typical temperature is closer to 60°C rather than 70°C. Therefore, it was unclear whether the risk extended to the temperature range typical of typical Western consumption.

Furthermore, it was necessary to separate the effect of temperature from the composition of the beverage itself. Coffee contains caffeine, polyphenols, and many other compounds, while tea has its own complex chemical composition. If the risk were determined by any component of a specific beverage, one would expect differences between tea- and coffee-drinking habits. In the new study, the direction of the association was similar regardless of the constituent of the hot beverage.

This is why the authors interpret the results primarily in terms of thermal stress on the mucosa. Higher temperatures, greater daily contact, and confirmation of the association across very different cultures combine to form a picture in which the specific fluid is less important than the repeated exposure of heat to the esophagus.

The study combined two large British cohorts

The first set of data came from the Million Women Study, a British study that initially included approximately 1.3 million women. The current analysis used 510,112 participants who provided information about hot drinks. The average age of the sample was approximately 64 years.

The second cohort is the UK Biobank, which includes both men and women. The analysis included 467,170 individuals, with an average age of approximately 56 years. Using the UK Biobank, the researchers were able to verify whether the observed pattern held true for other groups.

Participants in both cohorts were asked how they preferred to consume hot drinks: "warm," "hot," or "very hot." The number of cups of tea and coffee consumed per day was recorded separately, after which total consumption was divided into two categories: less than six hot drinks per day and at least six hot drinks per day.

The data were then linked to national cancer registries, hospitalizations, and mortality. In the Million Women Study, the average follow-up time was 14.2 years, while in the UK Biobank, it was 11.1 years. During this period, 2,348 new cases of esophageal cancer were identified: 1,045 squamous cell carcinomas and 1,303 adenocarcinomas.

Characteristic	Million Women Study	UK Biobank
Participants	510 112	467 170
Floor	women	men and women
Approximate average age	≈64 years old	≈56 years old
Average observation	14.2 years	11.1 years
Squamous cell carcinoma of the esophagus	799	246
Esophageal adenocarcinoma	617	686
They preferred "very hot" drinks	≈13%	≈17%
Drank ≥6 hot drinks/day	≈43%	≈44%

The hotter the drink, the higher the risk

In the initial analysis, without adjusting for the number of cups consumed, those who preferred "hot" drinks had a 1.79-fold higher risk of squamous cell carcinoma than those who preferred "warm" drinks. For "very hot" drinks, the risk was 3.47. The trend toward increased risk with increasing temperature was highly statistically significant: the p value for trend was less than 0.0001 in both cohorts and in the pooled analysis.

However, people who preferred very hot drinks were somewhat more likely to drink a large number of cups. Therefore, the authors separately adjusted for temperature by consumption frequency. After this adjustment, the association weakened slightly but remained very strong: the relative risk was 1.69 for hot drinks and 3.17 for very hot drinks.

Thus, the "threefold risk" cannot be explained simply by the fact that hot tea drinkers simply drank more tea. Temperature and quantity showed independent statistical associations. This is one of the most powerful aspects of the new study, as few previous major studies have been able to reliably separate these two factors.

The results were also consistent between the two cohorts. No statistically significant heterogeneity in the temperature effect was found between the Million Women Study and the UK Biobank. In other words, despite differences in age, gender composition, and recruitment methods, the direction and magnitude of the association were similar.

Temperature and squamous cell carcinoma of the esophagus

Preferred temperature	Relative risk before mutual adjustment	After taking into account the number of drinks
Warm	1.00	1.00
Hot	1.79	1.69
Very hot	3.47	3.17
Trend	p<0.0001	stable

The number of cups also had an independent meaning.

Almost half of the participants in both cohorts consumed at least six hot drinks per day. In the Million Women Study, this figure was approximately 43%, and in the UK Biobank, approximately 44%. This is particularly useful for statistical analysis, as the researchers had large groups of both moderate and very frequent drinkers.

After accounting for preferred temperature, drinking six or more hot beverages daily was associated with a relative risk of squamous cell carcinoma of 1.71 compared with drinking fewer than six. The 95 percent confidence interval was 1.51–1.94, so the likelihood that all of the observed difference was due to chance was small.

However, the researchers did not find a convincing statistical interaction between temperature and the number of drinks consumed. This means that the data do not support the claim that six cups of "very hot" tea confers any unique, supermultiplicative risk compared to the simple combination of the two factors. Rather, temperature and frequency contributed in parallel to the overall thermal load on the mucous membrane.

For adenocarcinoma, the situation was again much weaker. For six or more hot drinks, the relative risk was 1.19, while temperature itself was virtually unrelated to adenocarcinoma. Therefore, the main and most consistent finding of the study relates specifically to squamous cell carcinoma.

Why did squamous cell carcinoma respond, but adenocarcinoma did not?

The esophagus is lined with stratified squamous epithelium, which comes into direct contact with swallowed liquid. Squamous cell carcinoma develops from these cells and most often occurs in the upper two-thirds of the esophagus—the areas that are first exposed to hot liquids.

Adenocarcinoma has a different biology. It most often develops in the lower esophagus due to chronic gastroesophageal reflux and Barrett's metaplasia, when normal squamous epithelium is gradually replaced by glandular epithelium. Obesity, reflux, and other factors are particularly important for this type of tumor, while direct thermal stress may be less significant.

The authors view the increased risk specifically for squamous cell carcinoma and virtually absent for adenocarcinoma as indirect support for a thermal mechanism. If the result were due to a general unfavorable lifestyle or some component of coffee and tea, a less specific distribution between histological types would be expected.

This finding cannot be considered direct evidence of tissue damage, as participants did not undergo esophageal biopsies after each cup. However, it is consistent with animal experiments and laboratory studies, where repeated high temperatures damaged the epithelium, disrupted the mucosal barrier, and promoted inflammatory and precancerous changes.

How Repeated Heat May Promote Tumor Development

A single sip of very hot tea most often causes only an unpleasant burning sensation and a quickly healing superficial injury. The problem potentially arises with thousands of repeated exposures over many years. Each thermal injury forces the epithelium to activate repair and increased cell division.

The more frequently a cell divides, the greater the opportunity for genetic errors to become fixed, especially if other carcinogenic factors are present simultaneously. Ongoing damage can also sustain chronic inflammation, alter the tissue microenvironment, and temporarily disrupt the integrity of the mucosal barrier. These processes are considered to be the likely biological pathway between heat and tumor transformation.

Animal experiments support the plausibility of this mechanism. Repeated exposure to water at approximately 70°C in experimental models resulted in necrosis, epithelial hyperproliferation, and precancerous changes in the esophagus. Lower temperatures produced a significantly less pronounced effect.

However, heat likely does not act in isolation from other factors. Damaged mucous membranes could theoretically become more susceptible to carcinogens from tobacco smoke or alcohol. However, in the current study, the association with temperature persisted among non-smokers and people with low alcohol consumption, so the results cannot be explained solely by the interaction of these factors.

65 degrees is not a magical boundary between safe and dangerous.

One of the most important practical nuances of the study is that the researchers did not measure the actual temperature of the drinks. Participants independently selected the description "warm," "hot," or "very hot." Therefore, it's impossible to say that the "very hot" category precisely corresponded to 65, 70, or 75°C.

The International Agency for Research on Cancer uses a threshold above 65°C (145°F) to classify "very hot drinks," but the authors of the new study emphasize that this is a practical classification threshold, not an established biological threshold below which the risk suddenly disappears. Rather, thermal injury likely changes gradually with temperature, drinking speed, and the time the liquid is in contact with the mucous membrane.

Studies of consumer habits show that in the UK and US, tea and coffee are often consumed at around 60°C. Therefore, one of the most interesting interpretations of the new study is that risk differences may exist within the typical temperature range for Western populations, and not just at the extremely hot temperatures of around 70°C.

But it's impossible to determine a universal "safe temperature" from the article. A more reliable everyday guideline is to avoid drinking a beverage if it feels scalding hot in the mouth or requires careful microsips. The study shows the direction of the relationship, but doesn't pinpoint the exact degree at which individual risk begins to increase.

Adding milk did not explain the observed effect.

Adding cold milk clearly lowers the temperature of tea or coffee, so it could have been a hidden confounding factor. For example, if those who liked cooler tea almost always added milk, and milk itself somehow influenced cancer risk, the results would be more difficult to interpret.

The authors were able to test this question separately among 459,739 participants in the Million Women Study who reported whether they added milk to their drinks. About 73% added milk to tea and 64% to coffee.

However, the magnitude of the association between temperature and squamous cell carcinoma was not significantly affected by milk consumption. A statistical test of heterogeneity yielded p=0.20 for tea and p=0.32 for coffee, indicating no significant difference between groups.

This doesn't mean that cooling with milk is useless: it does physically reduce the temperature of the drink. The result suggests something else: the observed association cannot be explained solely by the groups' differences in milk consumption.

Smoking and alcohol also did not explain the threefold difference.

Smoking and alcohol are two well-established risk factors for esophageal squamous cell carcinoma. Therefore, it was especially important to rule out the possibility that people who prefer very hot drinks simply smoked more often or drank more alcohol, with temperature merely serving as a random marker for other behaviors.

Differences between groups did exist. For example, drinking large amounts of hot beverages was more common among smokers. Therefore, the authors consistently included potential confounding factors in their statistical models, including tobacco, alcohol, and a number of other characteristics. These adjustments only slightly altered the magnitude of the associations.

The association between temperature and cancer also persisted in analyses of never-smokers and people who consumed relatively little alcohol. The authors consider this to be an argument against the entire result being due to residual confounding from the two most important cancer-causing habits.

However, statistical adjustment is never equivalent to random assignment of participants. Other unaccounted factors may exist, such as dietary habits, drinking speed, mucous membrane condition, or cultural habits. Therefore, it is more accurate to say that major known confounding factors are unlikely to fully explain the results, rather than that they are completely excluded.

The effect persisted even after many years of observation.

The problem of reverse causality is particularly important for observational cancer studies. It's possible that a person is already developing early changes in the esophagus, causing them to perceive temperature differently, change the amount of drinks they drink, or choose cooler liquids. In this case, this unusual habit could be a consequence of the onset of the disease.

The researchers therefore specifically excluded the first five years of follow-up. If occult early tumors were the primary explanation, the association should have weakened significantly after removing early cases. This did not happen.

After five years of follow-up, the relative risk for "very hot" versus "warm" drinks was 3.16, virtually identical to the main analysis. For six or more drinks per day, the relative risk after five years was 1.78.

Furthermore, in the Million Women Study, habits were found to be relatively stable: when re-surveyed several years later, more than 60% of participants reported behaviors similar to those previously recorded. This mitigates, although does not eliminate, the problem of exposure being measured at only one primary point.

A threefold relative risk does not mean a high absolute risk

The number "3.17 times" sounds very significant, but it refers to the relative risk. Squamous cell carcinoma of the esophagus is relatively rare in the UK. Cancer Research UK estimates the lifetime probability of such a diagnosis at approximately 1%, so even a significant relative increase doesn't mean that one in three very hot tea drinkers will develop cancer.

A hypothetical example helps clarify the difference. If the baseline probability of a given event is 1%, a threefold relative risk would mathematically correspond to approximately 3%, not 300%. This is merely an illustration of the principle: an individual's absolute risk cannot be directly calculated from an article, as it depends on age, gender, smoking, alcohol, and other factors.

Furthermore, smoking and alcohol remain the most important preventable risk factors for squamous cell carcinoma. Cancer Research UK specifically emphasizes that quitting smoking and reducing alcohol consumption are more effective in preventing cancer than worrying about every cup of tea.

The temperature is interesting because it can be changed extremely easily and without negative consequences: just wait a few minutes. Therefore, even a relatively small contribution at the individual level potentially takes on significant social significance if millions of people regularly consume very hot drinks. This is precisely the principle behind the authors' calculation of a potentially preventable 14% of cases.

Where does the estimate "14% of cases are preventable" come from?

To estimate the population impact, the authors used the prevalence of very hot beverage preference in the UK Biobank and the magnitude of the observed association with squamous cell carcinoma. They then modeled a scenario in which all individuals in the "very hot" category shifted only to the "hot" category.

In this hypothetical model, the rate of preventable cases was approximately 14%. Based on national oesophageal squamous cell carcinoma incidence statistics, this equated to approximately 440 preventable diagnoses annually in the UK.

However, this estimate depends entirely on the assumption that the statistical association found is causal. If part of the difference is due to unaccounted for factors or subjectivity in self-reported temperature, the actual avoidable rate will be lower. Therefore, the authors explicitly formulate their calculation conditionally—"assuming a causal relationship."

On the other hand, the model assumes cooling only from "very hot" to "hot," not to "warm." Since the risk consistently decreased as the temperature category decreased, the authors suggest that more significant cooling could theoretically prevent even more cases. However, the current data do not allow us to determine the precise magnitude of such additional benefit.

What was modeled?	Grade
Original group	those who prefer "very hot" drinks
Behavior change	go to the "hot" category
Calculation conditions	the association must be causal
Potentially preventable proportion of squamous cell carcinoma	≈14%
Potentially preventable cases in the UK	≈440/year
Moving on to even cooler drinks	additional benefits are possible, but the exact amount is unknown

How the new work is stronger than previous studies

The main advantage is its size. Nearly one million participants and over a thousand confirmed cases of squamous cell carcinoma make this study the largest of its kind. This allowed us to separately evaluate squamous cell carcinoma and adenocarcinoma, something many smaller studies could not reliably accomplish.

The second advantage is the prospective design. Temperature and quantity of drinks were recorded before cancer diagnosis, and subsequent illnesses were identified through national medical registries. This is significantly more reliable than a retrospective survey of already-ill patients, who may have different memories of their past habits.

Third, reproducibility was demonstrated across two independent UK cohorts. The results from the Million Women Study and the UK Biobank were similar, despite differences in the participants. Furthermore, the direction of the effect is consistent with prospective studies from Iran, China, and Japan, as well as studies from the East African "belt" of esophageal squamous cell carcinoma.

Finally, the researchers separately tested several alternative explanations: number of cups consumed, smoking, alcohol, milk addition, the first five years of follow-up, and a wide range of demographic and behavioral subgroups. None of these controls eliminated the primary temperature signal.

But the study has fundamental limitations.

The most important limitation is that the temperature was subjective. One person might consider a drink at 58°C "very hot," while another might consider 65°C merely hot. Therefore, the results cannot be directly converted into a risk table for specific temperatures.

Second, temperature and quantity data were mostly recorded at a single point in time. Although the Million Women Study follow-up survey showed relatively high stability of habits, some people inevitably changed them over the course of ten or more years of observation. This measurement error typically blurs associations, but its impact is impossible to accurately assess.

Third, the study was observational. Cox models can adjust for smoking, alcohol, and other measured factors, but cannot completely eliminate unknown residual confounding. Therefore, independent experts characterized the result as a strong and consistent association, but not definitive causal proof.

Finally, the categories "warm," "hot," and "very hot" may refer to different absolute temperatures in the UK, Iran, or South America. This is why the authors caution against overly pinning risk on a single, universal 65°C mark and recommend research that directly measures the temperature of the beverage.

Limitation	Possible consequence
Temperature was reported by the participant himself	the exact temperature in °C is unknown
There is no direct measurement of each cup	it is impossible to set a precise safe threshold
Observational design	causality has not been definitively proven
There may be unaccounted factors	residual mixing
The main exposure is measured once	habits may have changed
UK population	the exact risk level may vary in other countries
65 °C is the conditional limit	is not a proven biological threshold

Should I cool my tea and coffee now?

From a practical standpoint, this is one of the rare risk factors that can be modified virtually without cost. There's no need to give up tea or coffee per se: the study didn't find that the composition of these beverages specifically increased the risk of squamous cell carcinoma. The association was primarily related to temperature.

The most sensible approach is to let the freshly brewed tea cool slightly and avoid drinking it if the first sips visibly burn your mouth. The authors don't provide a precise waiting time because cooling depends on the initial water temperature, the size of the cup, the volume of the tea, the room temperature, and whether milk is added.

There's no reason to interpret this result as a fear of ordinary warm coffee or tea. Independent experts emphasize that the absolute risk of squamous cell esophageal cancer in Western populations remains low, and beverage temperature is only one of many factors.

It's especially important not to let the new recommendation displace measures with a stronger evidence base. Not smoking and limiting alcohol consumption remain far more important preventative strategies for esophageal squamous cell carcinoma. Cooling drinks should be viewed as a simple additional step, not the primary way to prevent this disease.

The main conclusion of the study

In the largest British prospective study to date on beverage temperature, an analysis of nearly 977,000 people revealed a strong dose-dependent association with esophageal squamous cell carcinoma. After accounting for the number of cups consumed, "hot" drinks were associated with a 1.69-fold higher risk, and "very hot" drinks were associated with a 3.17-fold higher risk than warm drinks.

Quantity also had independent significance: six or more hot drinks per day were associated with a 71% increase in relative risk, even after adjusting for temperature. The associations persisted after accounting for smoking, alcohol, and other factors, and were consistent across individuals with different characteristics.

No similar temperature effect was observed for adenocarcinoma. This specificity, combined with experimental data, is consistent with the model that repeated thermal injury directly damages the esophageal squamous epithelium, forces it to constantly regenerate, and may eventually contribute to carcinogenesis.

But the study doesn't establish a precise, dangerous temperature or prove that switching from 70 to 60°C will reduce individual risk by a specific percentage. The most reliable practical conclusion is much simpler: there's no need to give up tea or coffee, but based on the current evidence, regularly drinking them scalding hot is wiser.

News source

Papier K., Gaitskell K., Floud S., Reeves GK. Hot Drinks and Oesophageal Cancer: Prospective Analyses in Around 1 Million UK Adults. International Journal of Cancer. 2026:1–9. The article was published online on 8 September 2026. DOI: 10.1002/ijc.70654.

This is an original prospective cohort study, not a randomized trial. The authors pooled data from 510,112 participants in the Million Women Study and 467,170 participants in the UK Biobank. Over a mean follow-up of 14.2 and 11.1 years, respectively, 1,045 cases of squamous cell carcinoma and 1,303 cases of esophageal adenocarcinoma were recorded.

The study was conducted by the Cancer Epidemiology Unit, Nuffield Department of Population Health, University of Oxford. The Million Women Study is funded by Cancer Research UK and the Medical Research Council UK; according to the article, the funding organizations had no role in the design, data collection, data analysis, or publication preparation. The authors declare no conflicts of interest.