People who said they preferred their tea or coffee very hot had more than three times the risk of developing esophageal squamous cell carcinoma as those who preferred their drinks warm, in a prospective analysis of 977,282 adults in two large UK cohorts. The association became stronger as reported drink temperature rose, while the same pattern was not seen for esophageal adenocarcinoma.
The researchers divided participants according to how they preferred to drink hot beverages: warm, hot or very hot. They also recorded how many cups of tea and coffee participants drank each day, grouping consumption into fewer than six or six or more hot drinks daily. The analysis then followed participants for cancer diagnoses using linked National Health Service records.
The distinction between the two major types of esophageal cancer was central to the analysis. Squamous cell carcinoma, or SCC, arises from squamous cells, while adenocarcinoma, or AC, is a different histological subtype. The researchers analyzed them separately because the two cancers have distinct aetiologies and could plausibly respond differently to repeated exposure to hot liquids.
Among the 977,282 participants, 510,112 came from the Million Women Study and 467,170 from UK Biobank. During an average follow-up of 14.2 years in the Million Women Study and 11.1 years in UK Biobank, the researchers recorded 2,348 incident esophageal cancers: 1,045 SCC cases and 1,303 AC cases.
For SCC, the risk rose steadily across the three temperature categories. Compared with participants who preferred warm drinks, those who preferred hot drinks had a pooled relative risk of 1.79, with a 95% confidence interval of 1.66 to 1.93. For those who preferred very hot drinks, the pooled relative risk was 3.47, with a 95% confidence interval of 3.07 to 3.91. The trend across temperature categories was highly statistically significant, with p<0.0001.
The pattern was consistent between the two cohorts. In the Million Women Study, the relative risks for hot and very hot drinks were 1.65 and 3.25, respectively, compared with warm drinks. In UK Biobank, the corresponding relative risks were 2.33 and 4.15. Statistical tests found no evidence that the size of the association differed materially between the two cohorts.
The picture changed when the researchers looked at adenocarcinoma. There was no evidence of an association between preferred drink temperature and AC in either cohort or in the pooled analysis. In the combined data, the relative risks for hot and very hot drinks compared with warm drinks were 0.93 and 0.95, respectively, with confidence intervals that included the possibility of no association.
That contrast between the two cancer types was one of the clearest features of the analysis. Higher preferred temperature was associated with SCC, but not materially with AC. The researchers said this pattern supports, but does not prove, the possibility that thermal exposure is involved specifically in the development of SCC.
Drinking more often was also associated with SCC
Temperature was not the only factor associated with SCC. The number of hot drinks consumed each day also showed a relationship with risk.
In the pooled analysis, participants drinking six or more hot drinks per day had a relative risk of 1.71 for SCC compared with those drinking fewer than six, after adjustment for preferred drink temperature. The 95% confidence interval was 1.51 to 1.94. The association with frequency therefore remained after accounting for the temperature at which participants preferred to drink their beverages.
The researchers also examined temperature and frequency together. Across the six combinations of temperature and daily intake, the highest SCC risks were seen among people who preferred very hot drinks and consumed six or more hot drinks per day. In the pooled analysis, that group had a relative risk of 3.21 compared with the reference group of people who preferred warm drinks and consumed fewer than six drinks daily. The corresponding confidence interval was 2.76 to 3.73.
The analysis did not find evidence that the association between the number of drinks and SCC depended on drink temperature. The test for interaction between temperature and number of drinks produced p=0.57 for SCC. For AC, the corresponding p value was 0.09.
For adenocarcinoma, the relationship with drink frequency was much smaller than the temperature association with SCC. After controlling for drink temperature, six or more hot drinks per day was associated with a pooled relative risk of 1.19 for AC, with a 95% confidence interval of 1.06 to 1.33.
The analysis covered two very large UK cohorts
The Million Women Study recruited about 1.3 million women in England and Scotland between 1996 and 2001 through the National Health Service Breast Screening Programme. Women were 50 to 64 years old when they received the initial questionnaire, which collected information on sociodemographic factors, body measurements, lifestyle and health. Participants were subsequently resurveyed at roughly three- to five-year intervals, while cancer diagnoses, hospital admissions and deaths were obtained through electronic linkage to NHS records.
UK Biobank recruited 503,317 men and women aged 37 to 73 between 2006 and 2010. Participants attended an in-person baseline assessment and provided information on lifestyle, sociodemographic factors and health. Their subsequent health outcomes were also obtained through electronic linkage to routinely collected NHS data.
For the current analysis, participants contributed follow-up time from baseline until cancer diagnosis, emigration, death or the end of available follow-up, whichever came first. The Million Women Study was followed through December 31, 2022. UK Biobank follow-up ended on December 31, 2020 in England, November 30 in Scotland and December 31, 2016 in Wales.
The researchers used Cox proportional hazards regression to estimate relative risks and 95% confidence intervals. Results from the two cohorts were then combined using inverse-variance weighting, while statistical tests assessed whether the estimates differed between the cohorts.
The participants’ reported drinking habits were widespread. In the Million Women Study, 43% reported consuming six or more hot drinks each day, compared with 44% in UK Biobank. About 13% of the women in the Million Women Study and 17% of UK Biobank participants preferred very hot drinks.
The temperature was self-reported, not measured
One important qualification is that the study did not measure the actual temperature of the tea or coffee participants drank.
Instead, participants reported whether they preferred their hot drinks warm, hot or very hot. The researchers therefore could not determine the absolute temperature corresponding to each category for an individual participant. They also noted that people’s perceptions of temperature can vary between individuals, populations and over time.
The study’s introduction notes that previous measurements have placed the temperature of tea and coffee commonly consumed in the United States and United Kingdom at around 60°C. The researchers said that further validation studies are needed to establish how the self-reported categories used in their analysis correspond to actual drinking temperatures.
That uncertainty also means the study does not establish a specific temperature at which risk begins to rise. The researchers pointed out that the 65°C threshold used by the International Agency for Research on Cancer to define “very hot beverages” is a pragmatic category for evaluating evidence rather than a demonstrated biological cutoff.
The researchers nevertheless observed a clear stepwise pattern across the self-reported temperature categories, with SCC risk increasing from warm to hot to very hot. They said this suggests that higher-risk drinking temperatures may occur below the temperatures cited in the IARC classification, although the actual temperatures represented by the categories would need further validation.
Other factors did not substantially alter the SCC association
Because smoking and alcohol consumption are relevant to esophageal cancer risk, the researchers tested whether differences in these and other characteristics might explain the temperature association.
Incremental adjustment for potential confounders did not materially change the associations of either drink temperature or the number of hot drinks with SCC. The temperature association also remained evident among never smokers and light drinkers.
The researchers examined the temperature association across 12 selected characteristics, including age, deprivation, education, exercise, smoking, alcohol consumption, height, body mass index, menopausal hormone therapy, parity, sex and the proportion of hot drinks coming from tea or coffee. The association between very hot and warm drinks remained substantially elevated across these subgroups, and there was no evidence of heterogeneity across the 12 characteristics for the temperature association.
For the association between six or more drinks and SCC, results were also consistent across the subgroups examined except for sex, where the heterogeneity test gave p=0.004. The researchers cautioned that this difference could be due to chance because the male results came only from UK Biobank and therefore involved substantially fewer cases.
The study also tested whether the associations might reflect reverse causation, in which early, undiagnosed disease could influence participants’ drinking habits. When the first five years of follow-up were excluded, the temperature association remained similar. For very hot versus warm drinks, the pooled relative risk was 3.85 for follow-up during the first five years and 3.16 for five or more years after baseline. For six or more versus fewer than six drinks per day, the corresponding relative risks were 1.40 and 1.78.
The researchers therefore said the similar results across follow-up periods suggested that the findings were not due to reverse causation, while acknowledging that preclinical symptoms could not be completely ruled out as a potential influence on baseline reporting.
Tea, coffee and milk produced similar patterns
The association with SCC did not appear to depend on whether participants mainly drank tea or coffee. The researchers reported similar associations for people whose hot drinks were primarily tea and those whose drinks were primarily coffee.
They also examined milk because adding milk could affect the temperature of tea or coffee and could potentially act as a confounding factor if milk consumption itself were related to cancer risk.
Among 459,739 Million Women Study participants who provided information about milk at the 12-year resurvey, 73% reported adding milk to tea and 64% reported adding milk to coffee. The researchers found little evidence that the temperature-associated risks differed according to whether women did or did not add milk to their tea or coffee. The heterogeneity p values were 0.20 for tea and 0.32 for coffee.
The authors interpreted these results as evidence against differences in milk intake explaining the temperature association. They also noted that the similarity between tea and coffee drinkers points toward beverage temperature rather than a particular drink constituent as a possible factor.
The proposed explanation involves repeated thermal injury
The study does not directly demonstrate how hot drinks might contribute to SCC. But the authors propose thermal injury as the most likely mechanism.
Their reasoning is based partly on the pattern of the epidemiological results. The association with temperature was confined to SCC rather than AC, and SCC typically arises from squamous cells in the upper two-thirds of the esophagus. The authors say those cells would be expected to be more susceptible to the effects of hot liquids.
The researchers also point to previous experimental evidence cited in the paper in which very hot water at 70°C promoted premalignant lesions and tissue injury in animal models, whereas lower-temperature beverages did not. They propose that repeated thermal injury could impair the integrity of the esophageal lining and promote carcinogenic processes.
This remains a proposed mechanism rather than something demonstrated directly by the UK cohort data. The present study measured participants’ reported drinking preferences and subsequent cancer diagnoses. It did not measure thermal injury to the esophagus or directly observe the biological process leading from hot drinks to cancer.
The estimated preventable fraction depends on a causal assumption
The researchers also used their findings to estimate what might happen if people who preferred very hot drinks instead consumed their drinks at the “hot” temperature category.
Assuming that the association between preferred drink temperature and SCC is causal, they estimated that 14% of esophageal SCC cases could be prevented if people who preferred very hot drinks instead consumed drinks at a hot temperature. Using national UK incidence rates for esophageal SCC, they translated that estimate into about 440 cases that could be prevented each year.
The researchers explicitly framed this estimate as conditional on a causal association. It is therefore not an observed number of cancers prevented by changing drinking behavior.
They also noted that the analysis found a graded increase in SCC risk across the reported temperature categories. On that basis, they said that reducing drinking temperature to something below the population’s average might potentially prevent a greater proportion of cases, but this too is a possible implication rather than a directly tested intervention.
The findings fit a broader pattern, but the UK data add a different temperature range
The authors compared their results with previous prospective studies from Iran, China and Japan, as well as an earlier UK Biobank analysis. Those studies also reported higher esophageal cancer or SCC risk among people consuming hotter beverages, although not all found an independent association with the amount consumed.
For example, a prospective study in Iran reported a relative risk of 2.41 for SCC among people who preferred very hot tea compared with those who preferred cold or lukewarm tea. A large China Kadoorie Biobank analysis reported higher esophageal cancer risks among men who drank tea daily and preferred increasingly hot temperatures. A Japanese cohort reported a 60% higher risk of esophageal cancer mortality among people drinking green tea at hot rather than moderate temperatures.
The researchers emphasize that many earlier studies were conducted in populations with different beverage practices and drinking temperatures. Their analysis instead examined tea and coffee consumption patterns commonly reported in the UK, providing evidence within a Western population in which hot drinks are widely consumed.
The consistency of the temperature association across the two UK cohorts was also reflected statistically. For SCC, the test comparing the cohort-specific estimates gave χ²=2.06 with p=0.37 for heterogeneity. For AC, χ²=0.46 with p=0.79.
Several uncertainties remain
The researchers identified several limitations beyond the lack of directly measured drink temperatures.
Drink temperature and intake were assessed at a single time point. However, information from a 12-year resurvey in the Million Women Study indicated that preferences were generally stable, with more than 60% of women reporting similar behavior to that recorded eight years earlier.
Even so, the authors acknowledged that self-reported preferences may not perfectly represent actual exposure. They also noted that subjective interpretations of terms such as “warm,” “hot” and “very hot” may vary among people and populations.
The prospective design reduces the possibility of biased recall caused by participants knowing they had developed esophageal cancer, but it does not completely eliminate the possibility that symptoms of disease that had not yet been diagnosed affected drinking behavior or reporting at baseline. The similar findings after excluding the first five years of follow-up provided some evidence against that explanation, but did not make it impossible.
Within those qualifications, the researchers concluded that both preferred drink temperature and the frequency of hot drink consumption were likely to be substantial risk factors for esophageal SCC, while neither showed a material association with AC. They said reducing drink temperature in populations where tea and coffee are frequently consumed could offer a means of SCC prevention.
The study was published in International Journal of Cancer.
