Putting out the fire - efficacy of common beverages in reducing oral burn from capsaicin

Abstract

Capsaicin is classically considered an irritant, due to the warming and burning sensations it elicits. Widespread consumption of chilis suggests many individuals enjoy this burn, but these sensations can be overwhelming if the burn is too intense. While substantial folklore exists on the ability of specific beverages to mitigate capsaicin burn, quantitative data to support these claims are generally lacking. Here, we systematically tested various beverages for their ability to reduce oral burn following consumption of capsaicin in tomato juice. Participants (n=72, 42 women, 30 men) rated the burn of 30 mL of spicy tomato juice on a general Labeled Magnitude Scale (gLMS) immediately after swallowing, and again every 10 sec for 2 min. On 7 of 8 trials, a test beverage (40 mL) was consumed after tomato juice was swallowed, including: skim milk, whole milk, seltzer water, Cherry Kool-Aid, non-alcoholic beer, cola, and water. Participants also answered questions regarding intake frequency and liking of spicy food. Initial burn of tomato juice alone was rated below “strong” but above “moderate” on a gLMS and continued to decay over the two minutes to a mean just above “weak”. All beverages significantly reduced the burn of the tomato juice. To quantify efficacy over time, area under the curve (AUC) values were calculated, and the largest reductions in burn were observed for whole milk, skim milk, and Kool-Aid. More work is needed to determine the mechanism(s) by which these beverages reduce burn (i.e., partitioning due to fat, binding by protein, or sucrose analgesia). Present data suggest milk is the best choice to mitigate burn, regardless of fat context, suggesting the presence of protein may be more relevant than lipid content.

INTRODUCTION

Consumption of spicy foods in the United States has grown substantially over the last decade [1], and new trends for spicy foods continue to emerge. Much of the heat in these dishes can be attributed to chili peppers (Capsicum solanaceae). The burn associated with chili peppers arises from several capsaicinoids, with capsaicin being the most abundant [2]. Other chemesthetic sensations evoked by capsaicin include warming and stinging [3], and in some individuals, it also elicits weak bitter sensations [4, 5]. Paradoxically, the burn from capsaicin can be (highly) desirable in humans, but the motivations behind the global popularity of chili peppers remains poorly understood. Potential factors that have been identified to date include thermoregulation, cultural traditions [6], personality traits [6, 7], and gender [6, 8]. (See [9] for additional detail). When overconsumption of pungent food occurs, there are numerous folk theories and myths about what beverage or food is best for mitigating the burn; however, there is minimal systematic data to support these widespread beliefs. Here, we focused exclusively on the effectiveness of different types of beverages, although it should also be noted that some other foods may also provide relief (see [10, 11]).

Various components of beverages may help mitigate the perceived burn of capsaicin through a variety of psychological, physiological, or physical mechanisms (or some combination thereof). For example, concomitant delivery of capsaicin with sucrose and rinsing with sucrose after capsaicin exposure each reduce oral burn. Sizer and colleagues [12] found that recognition thresholds for capsaicin were increased when varying amounts of sucrose were added (0.01M to 0.29M). A similar effect is also seen at suprathreshold concentrations: Nasrawi and Pangborn reported 40mM sucrose added to 2mg/L capsaicin reduces burn relative to the same capsaicin solution alone [13]. Those authors also found that rinsing with a 10% sucrose solution after exposure to 3mg/L capsaicin effectively reduced burn [14], suggesting beverages containing sucrose may be useful in mitigating excessive burn. The mechanism(s) behind such data remain unknown, but two possible explanations seem likely. One potentially related phenomenon that has been well investigated is mixture suppression of bitterness by sweetness [15–18]. In mixture suppression, the bitterness of a bitter compound presented in a mixture with a sweetener is less than the same concentration of the bitterant by itself. This interaction is perceptual in nature, and occurs centrally (see [19]); whether burn and sweetness also show central, perceptual interactions is unknown, but it seems feasible. Alternatively, another mechanism to potentially explain the reduction observed by Nasrawi and Pangborn would be analgesic effects of sucrose that occur via the opioid system [20, 21], although convincing evidence of sucrose analgesia in human adults is lacking.

Fat is another common food ingredient that possibly reduces oral burn following capsaicin exposure. It has been speculated that fat reduces oral burn via partitioning, due to the lipophilicity of capsaicin. That is, capsaicin may partition into the fat phase, thereby limiting the number of capsaicin molecules available for binding with the transient receptor potential cation channel subfamily V member 1 (TRPV1) receptor [22]. Increasing amounts of fat added to both cheese sauce and a starch paste containing capsaicin (0.4 to 1.3mg/L) reduced the burn relative to a no fat product [23]. Fat also appeared to be effective as a palate cleanser following rinsing with a solution containing capsaicin (0.5 to 8mg/L). When comparing different dairy products containing increasing amounts of milk fat (0.35, 19.68, and 39%) served with and without added sucrose (10%), the highest fat level with added sucrose was found to be the most effective at reducing burn [24].

In addition to prior work on the ability of sugar, fat, or fat/sugar mixtures to reduce burn, others have explored how temperature effects capsaicin burn. Nasrawi and Pangborn [14] investigated the effects of temperature, fat content and sucrose concentrations on reducing oral burn from capsaicin. Specifically, they tested water (5°C and 20°C) and milk based beverages with varying amounts of fat: skim (0%) and whole (3.5%) milk were presented without and with 10% sucrose [14].

Beyond sucrose, other tastants have also been reported to reduce capsaicin burn. Compared to rinsing with nothing, periodically rinsing with a solution containing either sucrose (0.3M), citric acid (0.0056M), sodium chloride (0.3M), and water were all effective at reducing the burn from 1 and 2mg/L capsaicin; notably, a quinine (0.0001M) solution, despite the somatosensory input from the water, behaved similarly to the no rinse trial [25]. Regarding threshold based measures, sodium chloride was ineffective at increasing the recognition thresholds of capsaicin [17]. At suprathreshold levels (2mg/L capsaicin), sodium chloride (0.3M), citric acid, (0.01M) and xanthan gum (0.2% wt/vol) were ineffective at reducing burn when added to capsaicin [13].

The present study builds on existing reports in the literature, as well as widespread folk wisdom, to explore the effectiveness of a diverse range of beverages to mitigate the oral burn of capsaicin. Here we selected seven common beverages to systematically test their effectiveness in reducing burn. They include skim milk, whole milk, sweetened fruit punch (Kool-Aid), non-alcoholic beer (O’Douls), a full sugar cola (Coca-Cola), carbonated water (seltzer), and room temperature water. The two milk products were chosen due to their differences in milk-fat content, which might potentially influence the partitioning of a lipophilic compound like capsaicin. Beer was chosen as it is a carbonated beverage frequently consumed with spicy foods, acknowledging that use of a non-alcoholic beer for logistical reasons might perform differently than a beer containing 4 to 6% ethanol. Kool-Aid and full sugar cola were chosen for their roughly similar sucrose content, although they differ in terms of the type of acid they contain, and the absence/presence of carbonation. Similarly, seltzer has carbonation like the cola, but no added sugar. Room temperature water was included to provide a secondary, alternative comparison to the no rinse condition. A priori, we hypothesized whole milk would be the most effective because it contains the highest fat content. Also, we expected that skim milk and Kool-Aid might also significantly reduce the burn due to the protein and sucrose content, respectively.

METHODS

Participants

Adults interested in participating a study consisting of one ~45 min laboratory visit completed an online screener before visiting the laboratory to ensure they met study inclusion criteria. These included being a nonsmoker between 18 and 55 years old, being not pregnant or breastfeeding, having no history of chronic mouth pain, no known smell or taste defects, not taking prescription pain medication (either recreationally or under care of a physician), no history of choking or difficulty swallowing, and no or lip/tongue/cheek piercings. Recruitment was not based on intake or liking of foods containing chili pepper, so both regular and infrequent consumers of chili peppers were enrolled.

A total of 76 adults participated, but data from 4 individuals were excluded due to issues with sample presentation or failure to comply with the study protocol, resulting in a final n of 72 (42 women, 30 men); their age ranged from 20 to 45 years with a mean of 28.3 (±7.8 SD). Ancestry was self-reported, with the majority of participants endorsing Caucasian ancestry (n=58), with low representation from Asian (n=9) and African/Black (n=1) ancestries; four participants self-identified as ‘other’. All procedures were exempted from full Institutional Review Board review by professional staff in the Penn State Office of Research Protections. Informed consent was obtained and participants were compensated with a small cash incentive.

Psychophysical scaling and practice sessions

A general Labeled Magnitude Scale (gLMS) was used to collect intensity ratings [26]. This scale ranges from ‘no sensation’ (NS) at o to ‘strongest imaginable sensation of any kind’ at 100; it was adjective labels of ‘weak’, ‘moderate’, ‘strong’ and ‘very strong’ at numerical values of 1.4, 6,17,35 and 51, respectively. All participants received instructions on the use of the scale prior to making any ratings [26]. These instructions included asking participants to not let affective responses (i.e., whether or not they like or dislike the sample) influence their intensity ratings; the instructions also encouraged participants to make ratings anywhere along the scale. After receiving these instructions, participants completed a 15-item warm-up/practice session, which included rating food and non-food items [27]. This practice helps participants use the scale in context to all sensations, and also allows for researchers too asses their performance and understanding of the use of the scale. All tests and data collection occurred via Compusense Cloud software (Guelph, ONT).

Stimuli and sampling procedure

Spicy tomato juice (Master of Mixes 5 Pepper Bloody Mary Mix; purchased locally from a retail store) was presented at room temperature in clear plastic cups; each cup contained 30mL. Test beverages included reverse osmosis (RO) water, cola (Coca-Cola Classic, purchased locally), Cherry Kool-aid (dry mix purchased at retail and reconstituted per package directions), seltzer water (store brand, purchased locally), skim milk (store brand, purchased locally), whole milk (storebrand, purchased locally), non-alcoholic beer (O’Doul’s; purchased at retail). All test beverages were chilled to refrigeration temperature (~4°C) except still RO water, which was served at room temperature (~20°C). Test beverages were served in 40mL aliquots in clear plastic cups labeled with three-digit blinding codes. All test beverages were pre-poured prior to testing, with the exception of the cola, seltzer and beer; these were poured immediately before the beginning of the trial for that beverage to minimize any loss of carbonation.

After the gLMS warm-up/practice, participants were instructed to consume 30mL of room temperature tomato juice. Immediately after swallowing, they rated the initial burn of the tomato juice on a gLMS. After the first rating, participants consumed 40mLs of the treatment beverage and continued to rate perceived burn every 10 seconds for 2 minutes. In total there were 8 trials: 7 included one of beverages mentioned above, and one trial had no test beverage. Order of treatments was counterbalanced across all participants using a Williams design. A minimum break of 2 minutes was enforced between trials to allow the burn to dissipate. Participants were instructed to not rinse with water within a trial, but were asked to rinse with water between each trial until they no longer experienced any sensation. Following the completion of all the trials, participants answered two questions: ‘How often do you consume spicy food?’ and ‘Do you like spicy food?’. Response options were [never, 1–3 times/month, 2 times/week, 3 or more times/week] and [yes, no, or no preference], respectively.

Statistical Analysis

Data were exported from Compusense Cloud (Guelph ONT) and preprocessed in Microsoft Excel (Redmond, WA). All statistical analyses were performed using SAS 9.4 (Cary, NC). Discrete interval time intensity data collected here were analyzed in three ways. First, repeated measure ANOVA was performed on raw gLMS ratings to test the effects of beverage type and time, while controlling for gender. Second, to summarize change over time into a single parameter, Area Under the Curve (AUC) values were calculated for each trial for each participant in Excel using the trapezoid rule (see [28]). These AUC values were then tested using repeated measures ANOVA to explore the effect of beverage type on burn. Finally, to quantitatively assess the amount of (potential) reduction in burn, data for each participant for each test beverage were normalized relative to their rating at time zero for that trial (i.e., the tomato juice burn rating prior to sipping the test beverage). Doing this provides a direct estimate of the amount of burn reduction for each beverage within a specific participant. To do this, a small constant (0.1) was added to all ratings to preclude potentially dividing by zero; proportional values at each timepoint were then obtained by dividing an individual’s rating for that beverage at that timepoint by their rating at t=0 for that trial. The normalized data were then tested using repeated measures ANOVA to test the effects of beverage type and time, while controlling for gender. All ANOVA models were performed via proc mixed using compound symmetry for the covariance structure; significant F-values were decomposed via Tukey HSD.

Results

Participant characteristics

Consumption and liking chili pepper containing foods varied substantially across participants. Slightly less than half of the participants indicated they consumed spicy food 2 or more times per month (n=30), while the remainder (n=42) reported less frequent consumption (‘1–3 times per month’ or ‘never’). Most reported liking spicy food (n=55), with 15 indicating no preference, and 2 indicating they did not like spicy food. Exact breakdowns by gender are shown in Table 1. There was no evidence of a relationship between gender and intake [chi square (3 df) = 3.22; p = 0.36] or preference [ chi square (2 df) = 1.77; p = 0.41].

Table 1:

Summary of spicy food intake and preference, stratified by gender

	Age (SEM)	‘How often do you consume spicy food?’ Frequency (%)				‘Do you like spicy food?’ Frequency (%)
		Never	1-3/month	2/week	3+/week	Yes	No	No Pref.
Women n=42	28.2 (±2.0)	6 (14.2)	19 (45.2)	13 (30.9)	4 (9.5)	30 (71.4)	1 (2.4)	11 (26.2)
Men n=30	28.5 (±1.5)	1 (3.3)	16 (53.3)	8 (26.7)	5 (16.7)	25 (83.3)	1 (3.3)	4 (13.3)
Total n=72	28.3 (±0.9)	7 (9.7)	35 (48.6)	21 (29.2)	9 (12.5)	55 (76.4)	2 (2.8)	15 (20.8)

All beverages tested reduced the burn of spicy tomato juice

Relative to control (i.e., the absence of any test beverage), all of the beverages significantly reduced oral burn. Means of raw (not normalized) ratings for oral burn on a gLMS are shown in Figure 1. In repeated measures ANOVA, main effects of condition (i.e., beverage type) [F(7,490)=75.68; p <0.001] and time [F(12,840)=165.61; p <0.001] were significant, and there was weak evidence of a possible gender effect [F(1,70)=6.04; p = 0.065]. Both two way interactions with gender - i.e., gender by time [F(12,840)=2.12; p = 0.014] and gender by beverage [F(7,490)=3.77; p = 0.0005] - were significant, while the beverage by time interaction was not [F(84,880)=0.81; p = 0.90]. In general, women rated the burn slightly higher than the men (−4.3 points on a gLMS). That is, in pairwise comparisons using t-tests, women had higher burn ratings compared to men for all beverages (p’s <0.03) except for cola and beer, which showed the same pattern without being significant (p’s of 0.12 and 0.06, respectively). Regarding the time by gender interaction, pairwise comparisons were significant for the first ~90s (again, with the women giving slightly higher ratings), while the last few timepoints were not, presumably due to a floor effect as ratings decayed toward zero. Finally, there was no evidence of three-way interaction.

Figure 1:

Least square means for perceived burn over time, controlling for gender. In ANOVA, burn decayed over time, and the beverages differed in their ability of reduce the burn of the spicy tomato juice (see text for details). Inset boxes on the right are magnified data from the beginning and end of the trial.

Given the apparent differences across beverages, we wished to better characterize how effective each beverage was at reducing burn over time. To do so, we tested AUC values via repeated measures ANOVA (see methods for details). As shown in Figure 2, the AUC values varied significantly across trials [F(7,497)=9.13; p <0.0001]. Initially, we planned to used Dunnett’s test to compare each beverage relative to the no rinse condition; unsurprisingly, this analysis indicated all 7 beverages resulted in significant reduction in AUC values relative to the no rinse condition (all p’s < 0.005; not shown). However, an anonymous reviewer suggested that we might instead consider the water rinse as the control, to account for non-specific flushing of unbound capsaicin from the oral cavity. Accordingly, we reran Dunnetf s test comparing all other trials to water. As shown in Figure 2, the no rinse condition was higher than room temperature water, and the whole milk, skim milk, and Kool-Aid were better than water at reducing the burn.

Figure 2:

Area under the curve (AUC) values calculated from burn ratings over time, compared via ANOVA. P-values represent Dunnett’s test versus water (in gray). Relative to room temperature water, AUCs for the no rinse condition were higher and AUCs for Kool-Aid, skim milk, and whole milk were lower.

Comparing the reduction in burn using normalized responses

To facilitate relative comparisons across beverages, and more importantly, to provide a direct estimate of the amount of burn reduction achieved by each, ratings were normalized and reanalyzed. This was done by expressing an individual’s burn rating at a given timepoint as a fraction of initial rating (t=0) for that specific trial. That is, for a single individual, a value of 0.8 at 10 seconds would indicate the burn they reported at 10 seconds was 80% of their rating at t=0 for that trial. Accordingly, a mean value of 0.5 for the group at 10 seconds would indicate that, on average, that specific beverage cut the burn of spicy tomato juice in half; likewise, a value of 0.2 at t=130 would indicate a reduction of ~80% after 2 minutes. Normalized values were then tested via repeated measures ANOVA.

As shown in Figure 3, burn decayed substantially over time [F(12,840)=no.30; p <0.0001], and the amount of reduction in burn caused by each beverage varied substantially [F(7,490) = 91.09; p <0.0001]. However, unlike the raw data above, after normalization relative to initial burn, there was no longer any evidence of a main effect of gender [F(1,70) = 1.83; p = 0.18]. Of the four possible interactions, three were not significant (p’s > 0.9). The only significant interaction was found for beverage by gender [F(7,490) = 4.50; p <0.0001], and examination of individual pairwise comparisons suggest this interaction was driven almost entirely by seltzer. Women showed significantly less reduction in burn by seltzer (p = 0.005) than men; none of the other beverages showed evidence of gender specific effects in burn reduction.

Figure 3:

Means for each timepoint and beverage expressed as a fraction of the initial value (t=0) for that specific trial. As with Figure 1, in ANOVA, burn decayed over time, and the beverages differed in their ability of reduce the burn of the spicy tomato juice (see text for details). Inset boxes on the left and right the magnified data at t=10 and t=130 seconds, respectively, with percentages indicating the fraction of initial intensity for that beverage and timepoint.

In Figure 3, inset boxes that zoom on values for t = 10 and t =130 have been added to highlight the relative amount of reduction in burn seen at the beginning and end of the trial. When burn ratings recorded immediately after swallowing the test beverage are examined (i.e., the inset on the left of Figure 3), it is clear that the greatest reduction in burn is seen for skim milk and whole milk, as their mean values at t = 10 are roughly half of ratings at t=0 for the same trial. Notably, this reduction is not merely a result of time, as average burn in the no rinse (control) trial at t =10 is slightly above 1, suggesting burn may otherwise continue to grow slightly if a rinse beverage is not provided. At the end of the trial, burn was substantially reduced (again, relative to initial burn) when skim milk and whole milk were consumed. Consistent with the AUC data described above, Kool-Aid also resulted in substantial burn reduction, although it was not quite as effective as the two milk samples.

Discussion

All beverages tested reduced burn over time relative to the no rinse control, but some were much more effective than others. However, when compared to room temperature water, only milk (skim or whole) and Kool-Aid outperformed water. The efficacy of milk in reducing burn, both immediately and over time, is generally consistent with prior reports [24]. It is widely assumed whole milk will be most effective, given capsaicin’s well-known hydrophobicity, as burn presumably drops with partitioning of capsaicin into the lipid phase (e.g., [22, 29]). Critically however, we failed to find evidence that whole milk (>3.25% fat) was more effective than skim milk (<0.5% fat). In pairwise comparisons at each timepoint (not shown), skim and whole never differed. Previously, Naswari and Pangborn reported “whole milk was more effective than skim milk in reducing mouth-burn” in a time-intensity study with 22 participants [14]. Strictly speaking however, their data do not actually support this claim, as they failed to observe significant differences in the ability of skim milk and whole milk to reduce burn. Thus, it seems critical to distinguish between fat content of a capsaicin containing food (e.g., [22, 29]) versus the fat content of a rinse agent after exposure (present data, and [14]). Regarding mechanism, present data suggest either i) the limited fat in skim milk is sufficient to partition capsaicin away from the receptor, or ii) the high efficacy of milk in reducing burn is due instead to another constituent in milk (e.g., casein or lactose). Recently, we observed protein binding reduces the taste of hydrophobic bitterants like quinine [30], so it is tempting to speculate that present results maybe due to protein binding rather than fat. Specifically, caseins are the primary protein in cow’s milk (~80%), so future work testing the ability of casein to solubilize or emulsify capsaicin, and to cut burn, are warranted.

Here, we also find that Kool-Aid - a fruit flavored drink containing citric acid and ~10% sucrose - effectively reduced burn relative to the two least effective beverages, cold carbonated seltzer and room temperature water (see Figure 2). This is consistent with numerous reports showing sucrose solutions can reduce burn [12–14,25, 31])· Notably however, the cola seems to be slightly less effective at reducing burn, despite containing roughly the same amount of added sucrose as Kool-Aid. That said, any apparent differences should be in interpreted cautiously. More generally, our data are consistent with the idea that sucrose reduces burn. Since sucrose analgesia is not robustly observed in human adults, it seems more likely that any cross-quality interaction is due to mixture suppression, similar to what happens with sweetness and bitterness in model systems (e.g., [16,19]) and real foods (e.g., [18]). Such an interaction may also contribute somewhat to the efficacy of the milk: although the overall sweetness of lactose is relatively low, it is present in milk.

Present data also align with prior reports on the interaction between chemesthesis and stimulus temperature. Solution temperatures below mouth temperature (~35°C) substantially reduce the burn of capsaicin, with the colder water being more effective [32]. Notably, during systematic manipulations of solution temperature, Green also noted that merely taking something into the mouth reduced burn, presumably via an inhibitory interaction with touch [32]. Although we did not explicitly or systematically manipulate mouth temperature per se, our data are consistent with inhibitory effects of both touch and temperature on burn, and we find that this appears to occur with real foods, not merely model systems.

Of the six cold beverages tested here, the three carbonated beverages - seltzer, cola, and non-alcoholic beer - were generally less effective at reducing burn (see Figure 3). Mechanistically, the tingle from carbonation arises from enzymatic conversion of carbonic dioxide into carbonic acid by carbonic anhydrase VI, and the intracellular acidification that results from this conversion causes responses via the TRPA₁ receptor [33] (reviewed by [34]). Recent data from Running [35] suggests plain carbonated water (i.e., seltzer) has more burn when sampled after a carbonated beverage containing ginger and chili extracts. Also, psychophysical data from model systems indicates that burning and tingling from carbonation increases as temperature drops [36], suggesting that use of cold seltzer, beer and cola here may have increased the chemesthetic sensations evoked by each [37]. Ethanol also activates the TRPV₁ receptor, but here we used non-alcohol beer for logistical reasons, so the question of whether the low amount of ethanol typically found in beer (~4 to 6%) might increase burn (via TRPV₁ activation), or decrease it (by partitioning hydrophobic capsaicin away from the receptor) remains an open question. Other data suggest the burn from ethanol is quite low at the concentrations found in beer [38], but we cannot rule out possible sensitization or synergy.

Multiple studies suggest differences in chili pepper consumption associates with differential ratings of burn in the laboratory, with frequent/heavy consumers reporting less burn relative to infrequent/non-users (e.g., [4, 39–41]). Accordingly, some heavy users tested here may have come into our laboratory with some unknown degree of desensitization. However, this does not meaningfully alter the present conclusions, as the analysis shown in Figure 3 uses each participant as his or her own control. There is no mechanistic reason to expect that the efficacy of the various beverages should differ between heavy and light users. Also, we observed a small but significant effect of gender on burn ratings in the raw (unnormalized) data. Reasons for this are unclear but one possibility relates to possible differences in oral cavity volume: if one assumes our women had smaller oral cavities than men (e.g., [42]), then the 30mL sample used here may have covered their oral surfaces more completely, resulting in more burn. Additional work would be needed to test this explanation.

Some limitations of the present work should be mentioned. Our focus here was on real beverages (versus model systems) that individuals might use to mitigate excessive capsaicin burn when they encounter nonoptimal doses in daily living; while this enhances the ecological validity of our data, it also precludes us from making strong inferences about specific mechanisms. For example, for cola, we cannot quantify the relative (and potentially offsetting) effects that arise from simultaneous influences of cold, touch, sweetness, and carbonation. Likewise, the beverages were served at temperatures they would typically be consumed; thus, our study cannot tease apart specific contributions of touch or temperature to the reduction of oral burn. Further work with experimentally formulated stimuli in a designed experiment would be required to do so. Also, our data were collected from a convenience sample on a university campus (i.e., a western, educated, industrialized, rich, democratic (WEIRD) cohort) so results may not generalize to other populations. Last, the spicy stimulus used here was a beverage. We fully expect our results would generalize to solid foods, but additional work would be needed to confirm this empirically. Nor did we test common semi-soft foods one might use to reduce burn (e.g., naan, yogurt, ice-cream), so we cannot say what food is the best, only which beverage is best among those tested here.

Conclusions

After exposure to commercially available spicy tomato juice that that evoked moderate to strong burning sensations, all rinse beverages significantly reduced capsaicin burn relative to the no rinse control. However, the seven beverages were not equally effective at reducing the burn of capsaicin. Sweetened fruit punch (i.e., Kool-Aid), and milk worked best, although full fat milk did not outperform skim milk, raising questions about the mechanism of action that makes milk so effective. These findings highlight a need for empirical testing versus logical inference from presumed mechanisms.

• Ability of common beverages to reduce oral burn from capsaicin was assessed

• All beverages tested reduced burn, but some clearly outperformed others

• Milk and fruit punch worked best, but whole milk was not better than skim milk

• Lack of a difference between whole and skim milk implies protein may play a role