|Image 1: Monkey milk tea by Conchibi. No matter how decorative it may look, according to various online sources, the milky monkey decoration is just about to annihilate the beneficial health effects of tea at the very moment it mingles with the antioxidant brew... or is this just another urban myth that is reaffirmed by 'gurus' all over the Internet?|
Answer Dr. Andro: This is one of the typical cases where just typing in a question in Google produces either answer A "tea with milk = no problem" or answer B "tea with milk = worthless swill" - it just depends on how you formulate the question - or maybe who on whether the milk industry advertises with Google? Who knows.... A more thorough investigation, on the other hand, turns up a whole host of studies, the results of which are similarly conflictive. So what can we do to find out the truth? Well, we will do our very own little meta review. Sounds pretty sophisticated, doesn't it? Certainly something only Dr. Andro can do. Well, not really... take my virtual hand and I will show you that, oftentimes, you can easily answer your questions yourself, if you mix PubMed, Excel and some of your own gray matter and simmer that for a few hours.
First we will have to find some real data related to the question
To "google" the answer would be a highly unreliable undertaking. Too many self-proclaimed experts (like me) run fancy looking blogs (like the SuppVersity) where they propagate factoids - something you will of course never find on the SuppVersity ;-) Our first task is thus to find real data, a process that obviously requires
- to know what real data is
- to know where to find this real data
- to know how to find this real data
- to have access to this real data
'Pubmeding' is 'googling' for smart people
www.pubmed.com) - and though I think that much of the dietary advice coming from the US National Institute of Health does more harm than good, the PubMed database comprises almost all medical papers that were published in any serious journal (not only the prestigious ones) from the realms of medicine, exercise, nutrition and beyond. On the PubMed website, we will type in the following search term "milk AND (coffee OR tea) AND (antioxidant OR polyphenol OR vitamin OR health)". It goes without saying that this will deliver all articles which deal with milk in relation to coffee or tea and their antioxidant, polyphenol, vitamin and caffeine content, as well as diverse health effects.
Note: You may have noticed that, out of personal curiosity, I have expanded our complex search term beyond Samir's original question to another often overlooked yet similarly healthy caffeinated beverage, of which the average American drinks 3.1 cups each day - coffee! And since I have heard rumors that milk could also influence the pharmacodynamics, i.e. the absorption and effect, of caffeine, when it is added to caffeinated beverages, I also included the term "caffeine" in our complex request. Notwithstanding the complexity of the latter, the use of multiple key words and tokens for their logical connection limits the number of "hits", i.e. studies that match our criteria to 503 (cf. image 1).From the initial 503 published articles the database has to offer, those papers which actually have our keywords in their title are obviously most likely to contain the information we are looking for. A mouse-click on the "see more" link below the "titles with your search terms" heading in the right column of the webpage (highlighted in image 2) reduces the results by -99.4% and leaves a manageable amount of three studies for us to 'review':
And, lo and behold, all three are actually relevant to the question we want to answer! Heureka! Our initial euphoria begins to fade, however, when we have a closer look at the titles. While the most recent study (Ryan. 2010) claims that the addition of milk reduces the total antioxidant capacity of black tea, Reddy et al. (Reddy. 2005) and Leenen et al. (Leenen. 2000) in their studies, five and ten years before, found no reduction in the antioxidant activity of (black) tea.... well, I guess the question would not have been worth being included in the famous *hem* Ask Dr. Andro column if typing in a few words into a search engine would have delivered the correct answer right away*.
1: Ryan L, Petit S. Addition of whole, semiskimmed, and skimmed bovine milk reduces the total antioxidant capacity of black tea. Nutr Res. 2010 Jan;30(1):14-20. PubMed PMID: 20116655.
Illustration 1: The way to a manageable
amount of reliable data for a meta-analysis
2: Reddy VC, Vidya Sagar GV, Sreeramulu D, Venu L, Raghunath M. Addition of milk does not alter the antioxidant activity of black tea. Ann Nutr Metab. 2005 May-Jun;49(3):189-95. Epub 2005 Jul 13. PubMed PMID: 16020939.
3: Leenen R, Roodenburg AJ, Tijburg LB, Wiseman SA. A single dose of tea with or without milk increases plasma antioxidant activity in humans. Eur J Clin Nutr. 2000 Jan;54(1):87-92. PubMed PMID: 10694777.
* We will have to postpone the effects on caffeine absorbtion and differences between the effects of milk in coffee vs. tea on another Ask Dr. Andro column, though, because the titles of our search results suggest that there won't be any information on these secondary issues in the respective fulltexts.
|Its good not to have to rely |
on pubmed for the links to
the fulltexts of papers you
are interested in; learn how
to find the FTs right at the
source, the webpage of the
Diging into it: Understanding the research design
Since we already know that the antioxidants in tea are good for us - otherwise we would hardly be worried that the addition of milk could "disturb or destroy" them, as Samir put it - the synopsis of the research, an obligatory part of the introduction to any scientific paper, which, in the case of the Ryan paper, focuses on the health benefits of tea and the tea consumption of the average UK citizen, is not really of interest to us. What is of utmost importance, though, is the research hypothesis, which is usually stated at the end of the introduction. Here, Ryan & Petit state three concrete objectives, which are...
- To analyze and compare the antioxidant capacity of five brands of commercially available tea
- To determine the effect of infusion time on the antioxidant capacity
- To examine the effect of different volumes of bovine milk (10, 15, and 20 mL) and milk of varying fat content (whole, semiskimmed, and skimmed) on the total antioxidant capacity of the 5 teas
Next, we are going to collect relevant information on the actual experimental procedure
- the tea was bought at a conventional supermarket
- content of the tea bags varied slightly (2.7-3.3g) between the 5 commercial brands
- 200ml of 90° hot water were used for infusion
- 6 infusions for all 6 infusion-times (1, 2, 3, 4, 5 and 10 min); thus 36 individual samples for each brand
- individual analysis of concentration effect of squeezing (10s right after removal) the tea bag after 1, 2, 4 and 8min of brewing; another 24 samples for each brand
- addition of 10, 15 and 20ml whole, skimmed and semi-skimmed milk (after cooling the sample by infusion of the same amount of cool water); another 54 samples for each brand
- technique used to analyze the 570 samples: ferric iron reducing antioxidant power (FRAP) assay
Finding and analyzing the relevant data
While obviously, the major results of this large scale investigation can be found in the text, the tables and figures the authors provide are of even greater value to independent thinkers like us. Using some sort of spreadsheet software we make good use of the 'raw' data on the influence of infusion times on the FRAP concentration of the freshly brewed tea. Instead of relying on the confusingly overcrowded and not very informative figure the authors provide, we create our own figure (figure 1) which comprises the FRAP values of tea 1 (median FRAP) and tea 2 (maximal FRAP) after 1, 2, 3, 4, 5 and 10min brewing with and without a tea bag. In that, we go beyond the simple graph Ryan and Petit had to offer, and bring to light how meaningful the negative effect of the tea bag in terms of the total antioxidant concentration of the brew actually is.
|Figure 1: Although authors usually provide you with graphical illustrations of their data, it is often worth-wile to plot your own graphs from the data they provide, to illustrate and understand things the authors may have considered ir-/less relevant and have thus not included in their graphical analysis - in this case this would be the effect of the tea bag, which is missing from figure 1 in the original study by Ryan & Petit (data from Ryan. 2010)|
Just because we are too lazy to tinker with the loose tea leafs we give away somewhere between 50% and 5% of the total antioxidant capacity (depending on the brand and infusion time of the tea).In that, it is interesting to note that the overall effect appears to be much greater with Tea 1 - as it can be seen in figure 1, the FRAP values of teas 1 and 2 are almost identical, if they are removed from the tea bag before infusion, which suggests that the quality of the teabag is an unexpected, yet important factor, when it comes to brewing an antioxidant rich tea. Altogether, these observations tell us that the use of teabags is generally not advisable, if you do not like the bitter taste long-infused teas tend to develop.
Doing some basic math to be able to estimate effect sizes
As interesting (and surprising) as they may be, the insights on the effects of tea bags on the amount of antioxidants are not really relevant for our personal research question, which relates to the effect of milk on the antioxidant activity of tea. Now, before we even bother with the data analysis, imagine, we did not even have data on the antioxidant activity of the tea + milk mixture. In that case, the only, yet questionable way to decide whether the addition of milk reduces the anti-oxidant value of tea would be to analyze the antioxidant value of milk, which is something Ryan & Petit actually did, and do some basic volume-based calculations.
Ryan & Petit found that fresh whole milk, semi-skimmed and skimmed milk had about 68%, 66% and 60% of the maximal FRAP value of (unbagged) tea and that these values declined by roughly 37.5% and 13.75% after 7 and 14days, respectively. Now, if we add a liquid with a FRAP value of say 5000µmol/L (fresh whole milk) to a brew of tea with a frap value of >8000µmol/L (Tea 1, leafs only) we will obviously reduce the total antioxidant level of the mixture. If we simply assume that the pre-cooled tea (20ml cold water added) has a FRAP value of 8000µmol/L (the exact reduction due to the addition of water would only matter, if the author had not used the 'cooled' specimen as reference, anyway), the addition of another 20ml of whole milk with a FRAP of 5000µmol/L would create a 11:1 mixture of high vs. low FRAP liquids, which would then have a FRAP value of 7.750µmol/L, which would obviously be a -3.125% reduction in the total antioxidant capacity of our tea by the addition of whole milk.
The smart SuppVersity teacher and students that we are, we are without doubt aware that this basic calculation oversimplifies our "problem". Yet, imagine the scientists found that the addition of 20ml of whole milk did in fact reduce the FRAP value by no more than -3.125%. In that case, we would have a reduction, but that reduction would be meaningless in view of our research question, because overall you would even get more antioxidants from your 240ml of tea with added milk and water than from your 220ml of tea without milk.
|Figure 2: Relative changes in FRAP values of Tea 1 due to milk addition compared to the addition of the same amount of water (data calculated based on Ryan. 2010)|
Understanding the results of the study
If we think about the composition of whole, skimmed and semi-skimmed milk two things which are closely related to the fat content come to mind: First, low fat milk contains more carbohydrates than full fat milk. And, secondly, the low fat variety also has more protein on a liter per liter base than its full fat precursor. On the other hand, previous studies Ryan & Petit mention in their discussion found a profound interaction of tea polyphenols and protein.
Polyphenols can not only bind to proteins, but that the "interaction between flavonoids and proteins affects their antioxidant capacity in vitro".If you are interested in optics you will also know that skimmed milk derives its bluish tint from the absence of fat globules, so that only the rayleigh-scattered high frequency blue light from the casein proteins hits the retina of your eye. If you combine that knowledge with the findings of Luck et al. (Luck 1994) who found that proline-rich proteins such as casein have a very high binding affinity for polyphenols, the answer to the influence of the macronutrient composition of the different types of milk is apparent: The casein fraction which is particularly high in skimmed milk binds to the beneficial tea polyphenols and thus decreases its antioxidant capacity. Well, I guess this would be our answer, then: Adding milk to your tea is not advisable, because its protein content will bind the beneficial polyphenols. And yes, it would be, if there were not the two other studies our initial PubMed search had brought up, which report no effect of milk on the total antioxidant capacity of black tea?
Is the newer study always right? Not one study is like another.
The easiest way to tackle the conflicting data would be to assume that the 2010 study is likely to be more accurate than the 5 and 10 years older studies by Reddy and Leenen, but would that be scientifically valid? I don't think so. So what do we do? We take a closer look at the methodology of the Reddy and Leenen studies... it won't take us long to find that in contrast to the Ryan & Petit, both Reddy and Leenen analyzed serum samples of subjects after ingestion of the respective tea+water / tea+milk mixtures. This major, yet easily overlooked difference is of great importance, because it does not take a rocket scientist to know that one of the fundamental functions of our digestive tract is to disassemble the food we consume so that the cells in the endothelial lining of our intestines can absorb the nutrients and deliver them to our bloodstream. Analyzing the ferric reducing ability of human plasma after the subjects consumed a tea + water / tea + milk mixture is thus completely different from putting the mixtures themselves to test.
With the influence of digestion, it is quite obvious that both the ingestion time, or rather the status of the digestive tract at that moment the subjects consume the samples, as well as the time at which the blood for the analysis is drawn post ingestion of the teas is of paramount importance. In both the Reddy and the Leenen study, the subjects reported to the laboratory fasted, and the studies also share some problems in regard of the tea preparation.
- Reddy, for example used only 280ml of water for the preparation of the tea to which then 70ml milk would be added. Since there is however a saturation level for the maximal amount of antioxidants a unit volume of water is able to hold, the 'no milk reference sample', which had been boiled in 350ml of water, probably had an (unfortunately unquantifiable) 'antioxidant advantage' over the 'milk sample'.
- In the Leenen study, on the other hand, the milk was just added to an accordingly reduced amount of tea (60ml milk to 240ml tea vs. 300ml tea), which does also reduce the overall amount of tea polyphenols, the subjects in the tea + milk groups were consuming.
So what? Does milk destroy the antioxidant value of tea?
Unfortunately, we cannot be sure that the aforementioned methodological crudities in the studies by Reddy and Leenen can adequately explain the milk-induced reductions in total antioxidant uptake. The -23% reduction in total ferric to ferric ion reduction capacity Leenen et al. measured in their study does yet match the reduced intake of tea polyphenols in the 240ml tea + 60ml milk vs. the 300ml tea group (-20% less polyphenols) pretty well. In view of the fact that Leenen et al. also used 2.5 times more whole milk (20% milk in tea vs. 8% milk in tea) than Ryan & Petit in their in-vitro study, one could do another ignorant calculation in order to estimate how effective our digestive system cleaves the proteins from the polyphenols:
5.73% of the total polyphenol content of 220ml tea are bound by the addition of 20ml whole milk; thus 20ml whole milk can bind the polyphenols from 12.606 ml of tea and 60ml of whole milk should be able to bind the polyphenols from 37.818ml of tea, so that the subjects in the Leenen study who consumed their tea with milk should be missing an additional 15.7575% of the antioxidants, not just meager 3%, which would ultimately mean that their digestive tract recovered >80% of the protein bound antioxidants from the milky brew. [note - this calculation ignores that the milk would have an antioxidant value by itself, as well]No matter how "scientific" these funky figures with 4 decimal places may appear - in essence the whole calculation did not only oversimplify the matter to a degree bordering bloggerish pseudoscience. It assumes, for example, a linear binding affinity, i.e. more milk = more binding, which obviously goes against what Ryan and Petit have found (cf. fig. 2). More importantly, though, it was also completely unnecessary in view of the fact that the 'missing' 3% we tried to explain are statistically non-significant, because standard error of the mean AUC (area under the FRAP curve) we based our calculations on is 19.64% and thus more than 6x above the loss we were going to lose our sleep over.
Before we go into even more speculative, or should I say outragous (?) calculations on the influence of the ratio of tea to water on the total amount of antioxidants present in tea after 3 minutes of infusion, in order to do similarly meaningless calculations for the Reddy study, we better bring this highly educative issue of our (almost) weekly Ask Dr. Andro column to a close and summarize our results as follows:
- the protein fraction in milk can bind antioxidant molecules in tea
- the antioxidant value of the tea + milk mix is reduced by up to -18% depending on the type and amount of milk you are using
- skimmed, i.e. low fat, 'high' protein milk reduces binds more antioxidants than whole milk
- the effect of milk on polyphenol binding in tea is non-linear or, in other words, if you double the amount of milk you won't lose twice as much of the antioxidants
- in in-vivo studies, i.e. studies that analyze the sera of living human beings after consumption of tea and tea + milk samples, the increase in serum FRAP (Ferric reducing ability of plasma) has been found slightly reduced by the addition of relatively high amounts (20%) of milk, but...
- this did not compromise the desired positive downstream effects on health markers measured in the Reddy study and
- the lower FRAP values after ingestion of the tea + milk mixtures could be reducible to the individual study design