Wine and health — why is there so much argument, pro and con?

We all know that every second media report extolls the health benefits of wine, while every alternate one warns us of the evils of wine (see Health effects of wine). Even the federal Dietary Guidelines for Americans seems unsure about the matter. This leads to confusion among the populace, even engendering cynicism. However, leaving aside the idea that the various authors might have their own agenda when writing their articles, this situation is not actually unexpected.

From my own perspective as a scientist in biology, there are three very obvious reasons why media reports about human health might contradict each other. This has everything to do with medical experiments, which are supposed to be the basis for the claims both pro and con. I don’t necessarily condone the experiments themselves, but I can see why their outcome might be confusing.

First, biology is complex, which may sound trite, but it contains the essence of the problem. Not only is biology more complex than we imagine, it frequently turns out to be more complex than we can imagine. The current Covid-19 pandemic is a perfect case in point — the more we learn about the SARS-CoV2 virus, the more we realize just how complex is the human immune response to this virus.

Remember, biology is actually the study of why we are not dead. The natural state of the universe is for things to be not alive, and yet lots of things are actually alive, at least for a while. Being alive turns out to be much more complex than being dead — being inanimate does not require very much, really, but being animate has all sorts of complex chemical requirements.

For our purposes here, the important consequence of this complexity is that anything with positive effects on one part of our body is likely to have at least some negative effects on some other body part; and vice versa. Nothing is ever “all good” or “all bad” in biology, no matter how much optimism might encourage you towards a search for the former. So, it is rather easy to find circumstances where wine seems to have positive effects on health, and also just as easy to find the opposite. It would be naïve to expect anything else.

Experiments

The second of our three reasons is that it is impractical to actually carry out the medical experiments required to study human health. For example, there is much discussion at the moment about the effectiveness of having the public wear medical-style face-masks during the current pandemic (eg. “masks might help you”). This is because, as far as I know, no-one has ever done an experiment to find out whether they do help in practice. It would not be easy! The same thing applies to understanding transmission of the virus from person to person (Scientists estimate the speed and distance of coronavirus transmission when people cough, sneeze, speak — and run).

The classic form of a scientific experiment is what we call a Manipulative Experiment (or Intervention Experiment). Here, we try to deliberately modify the world in some way, in order to test some hypothesis about how the world functions. This is often straightforward for inanimate objects, at least in theory, but it is almost impossible for animate objects.

For example, to study dietary health, we would need to take two groups of people, one lot healthy and the other lot unhealthy, and subject each group to both an apparently healthy diet and an unhealthy one. That is, we try to make some of the healthy people healthy, and make some of the others unhealthy; and we also try to make some of the unhealthy people healthy, and make some of the others unhealthy. That gives us four different experimental groups of people to study.

How are we going to get people to agree to do this? After all, we have to assign the people to the different treatments (making them healthy or unhealthy) at random — the people cannot choose which treatment they will get. You’d have to be a nutter to agree to be made unhealthy when you are currently healthy, even in the name of Science. So, the ideal medical experiment is called a Randomized Controlled Trial, and you don’t see too many of them when studying healthy vs. unhealthy lifestyles.

To take one straightforward example, let’s look at a particular experiment studying the hypothesized health benefits of a Mediterranean diet (with wine!):

Tarini Shankar Ghosh, and 30 others (2020) Mediterranean diet intervention alters the gut microbiome in older people reducing frailty and improving health status. Gut 69: 1218–1228.

They describe their experiment this way:

We profiled the gut microbiota in 612 subjects across five European countries (UK, France, Netherlands, Italy and Poland) before and after the administration of a 12-month long Mediterranean Diet tailored to elderly subjects.

This is all very well (people before and after changing to a Mediterranean diet), but where is the rest of the experiment? Where are the people who normally consume a Mediterranean experiment but changed to another diet? The Mediterranean diet is characterized by increased consumption of vegetables, legumes, fruits, nuts, olive oil and fish (and maybe a glass of wine!), along with low consumption of red meat, dairy products and saturated fats. Well, for a proper experiment, some fish-and-vegetable eaters should have started eating a meats-dairy-fats diet, and also been measured before and after their change. This gives us four groups of data in our experiment, not two, as shown in this figure.

Sadly, this failure is all too typical of medical experiments.

Finally, even if some people do agree to take part in an experiment, it is unethical to put people in experiments that can have risky outcomes. The medical scientists aren’t going to jail just to find out in which ways wine is good for you! Besides, the sort of person who volunteers to take risks is rarely a representative example of Homo sapiens.

So, most medical experiments are what we call Descriptive Experiments (or Observational Experiments). Here, we don’t try to manipulate people’s health, but instead study different groups through time, to see how things turn out under different natural circumstances. For example, groups of people might be studied who drink different amounts of wine in their daily lives, and we try to detect some health differences between the groups through time.

This can be very helpful way to learn things, but it is not necessarily a good substitute for a Manipulative Experiment. The main limitation is that we might end up with a biased sample of people, so that our results are skewed in some (unknown) way. That is, other factors can get in the way of studying the thing you want to be looking at — these are called confounding factors. For example, there is an obvious association between health and wealth — wealthy people can afford to look after themselves properly. Moreover, wealth is also associated with drinking — heavy drinkers and abstainers tend to be poorer than moderate drinkers. So, how do we extricate the effects of wealth on health compared to the effects of drinking? This question does not arise for Manipulative Studies, because we randomly assign people to the experimental groups.

Time

The third of our three reasons for apparently contradictory medical results is that human health is a long-term issue, while experiments don’t usually get funded for all that long. Most experiments on humans end before there has been much effect on the participants’ health. Indeed, it can take decades to find out about health issues; and even the longevity of of the scientists themselves can cut short such experiments!

To make this point clear, let’s look at an actual example, where we can also comment on the first two points above, as well:

Katherine M. Keyes, and 6 others (2019) Alcohol consumption in later life and mortality in the United States: results from nine waves of the Health and Retirement Study. Alcoholism: Clinical & Experimental Research 43: 1734–1746.

This is a Descriptive Experiment, since there is no actual medical intervention. A large group of Americans were simply followed through time, being evaluated every couple of years:

We report on 16 years of follow-up from the Health and Retirement Study (HRS) cohorts born 1931 to 1941 (N = 7,904, baseline mean age = 61, SD = 3.18). Respondents were queried about drinking frequency/quantity. Mortality was established via exit interviews and confirmed with the national death index.

The idea was to see whether drinking habits have any effect on how long we live. In practice, we cannot do a Manipulative Experiment, because that would involve (among other things) getting some abstainers to take up alcohol, and getting some alcoholics to go cold turkey. It is much simpler to just let elderly people drink whatever they want, and then see how long they live. If we have a large enough group of people, and we continue to follow them for long enough, we will eventually get an answer to our experimental question.

The problem is that lots of factors in a person’s life affect how much they drink, and some of these also directly affect their health. There are also factors that affect health indirectly. Even worse, some of these factors vary throughout our lives, not necessarily remaining constant — as we get older we change our behavior, especially in response to illness. Therefore, these other factors can confound our experiment, which is aimed to study the effects of alcohol alone. This is the basic limitation of Descriptive Experiments (reason 2 above), especially when studying complex phenomena (Reason 1 above).

The authors explicitly try to address this issue:

Time-varying confounders included but were not limited to household assets, smoking, body mass index, health / functioning, depression, chronic disease; time-invariant confounders included baseline age, education, sex, and race.

One of the interesting issues here is reverse causation. For example, people might reduce alcohol consumption due to illness, making it look like smaller amounts of alcohol cause you to die earlier.

The important aspect of the experimental design was to distinguishes moderate drinking from abuse drinking. The failure to do this is repeatedly discussed in wine blogs (eg. One more reason to be wary of alcohol health studies). One of the comments to that Wine Curmudgeon post notes; “temperance movements focus on the deleterious effects of excessive alcohol consumption and routinely fail to acknowledge the beneficial effects of moderate consumption.” An “all or none” approach to health management is ludicrous, biologically (but temperance arguments have never been biological, they are instead social).

So, the authors grouped the participants into these categories, based on drinking habits:

• Lifetime Abstainer — <12 drinks in their lifetime
• Current Abstainer — drank in the past but not during the experiment
• Occasional Drinker — drinking less than 1 day per week
• Moderate Drinker — 1-2 (women) or 1-3 (men) drinks 1 or more days per week
• Heavy Drinker — >3 drinks per day or >5 drinks in a single session for men, or 2 resp. 4 for women.

The results for males and females are shown in this pair of graphs, which show what proportion of people were still alive at each time (click to enlarge).

Note, first, that it took at least 3 years to find a detectable difference between the groups for men, and 5 years for women; and most of the people are still alive at the end of the measurement period, so that there is a long way to go in this experiment. That is the main take-away message here — any shorter experiment will be misleading.

However, we can also note that if we take the Lifetime Abstainers as our comparison for longevity, the Current Abstainers and Heavy Drinkers do worse (ie. tend to die sooner) while the Occasional Drinkers and Moderate Drinkers do better (ie. tend to live longer). The result for the Heavy Drinkers needs no explanation; but the authors note that the Current Abstainers were often people who had quit drinking due to illness, so that lack of alcohol was not directly associated with them dying sooner.

This experiment lead the authors to conclude that: “There are consistent associations between moderate / occasional drinking and lower mortality.” However, there was less of an effect for men versus women, and for smokers versus non-smokers (ie. smoking will eventually get you, whether you drink or not).

Conclusion

So, there you have it — three good reasons why reports of the outcome of medical experiments might contradict each other. Does moderate wine consumption improve your health? In some ways, apparently “yes”; but don’t overdo your drinking — after all, there is always this thing called “too much”.

The increasing cost of government compliance for small wineries

In my last post (How large does a winery have to be, to be consistently profitable?), I noted that there seems to be a minimum sales-turnover size below which wineries struggle to be profitable. However, there are also other factors that seem to preferentially disadvantage small wineries, compared to larger ones. One of these is the financial cost of government regulatory compliance.

For example, in the Napa Valley, the Save the Family Farms group of grape-growers note that, even if a winery produces less than 5,000 gallons of wine per year (eg. 1,000 cases), current Napa County regulations require expenditure that is no different from those of much larger wineries. Under these circumstances, the smaller wineries are preferentially disadvantaged.

This seems to be a general problem, and may indeed be increasing.

We can see this by looking at the annual Wine Industry Benchmarking and Insights survey in New Zealand, produced by Deloitte organisation, in conjunction with the ANZ Bank and the New Zealand Winegrowers association. I used this information in my previous post, which you can check for more explanation.

Each year since 2006, a survey questionnaire has been sent to all members of New Zealand Winegrowers, asking about details of their previous year’s financial statement. Each report compiles the response data on financial aspects like supply and demand, revenue streams, profitability, equity, and return on assets, as well as other important things like key markets and customer connections.

The reports also subdivide the wineries into four size tiers, based on total annual revenue in $NZ: $0-$1.5 m, $1.5-$5 m, $5-$10 m, $10-$20 m, and $20 m+. [Note: $US 1 ≈ $NZ 1.5]; and aggregate data are presented for each of these tiers.

With the apparent exception of the current report, one section of each report concerns Issues and Challenges. Each year, up to 11 issues were ranked by the participants in order of their perceived importance. These issues will seem all too familiar to any winery:

• Grape supply (too little) or affordability of land
• Grape supply (too much)
• Labour supply / cost
• Terms of trade (cash cycle)
• Sales margin pressure
• Access and/or cost of capital, including interest rates
• Government and other compliance costs
• Company tax rates
• Distribution, including marketing product overseas
• Exchange rates
• Succession

Some of these issues were ranked as very important by all winery sizes, for example Sales margin pressure, and Distribution. However, others differed somewhat, notably Government and other compliance costs. Above is a graph of the average ranked results from the 2017 survey (click to enlarge). Note that for the smallest-size tier of wineries, the top three issues were roughly equal in importance. Also, note that for the wineries in second-smallest tier chose only the same top two issues, and differed notably on the other issues,

Since our interest in this post is on the costs of government regulatory compliance, it is interesting to note the change through time in the ranking by the smallest wineries of the issue Government and other compliance costs. The next graph illustrates this, where rank 1 = most important issue. The importance of government compliance has thus regularly increased in relative importance recently, so that it is now in the top 3.

The Benchmarking reports offer no insight into the exact cause of this rise in importance for the smallest wineries, especially given that the rankings for the second-smallest tier remained between between 6–10 over the same time period. Indeed, the reports focus very much on issues related to wine exports, which is not unexpected given the New Zealand wine industry’s dependence on foreign income (the domestic market is small relative to production).

However, the reports do note that generally there is no “one size fits all” approach for wineries. Sadly, this fact is rarely acknowledged by government regulations.

Such regulations can include, but are not restricted to, aspects of vine growth (including biodiversity management, and sustainable agriculture), winemaking (including food safety, and renewable energy usage), human management (including on-site facilities for customers), and business standards (including local and export marketing). For their members, New Zealand Winegrowers has a document recommending Sustainable Winegrowing New Zealand: Standards, which lists acceptable practices that meet government regulations in each of these areas. No mention is made of the financial cost of implementing any of these practices.

How large does a winery have to be, to be consistently profitable?

This question is of great interest at the moment. The Covid-19 pandemic has, to coin a phrase, hammered much of the wine industry. In the USA, for example, there has a been a large spike in wine sales for supermarkets and large online retailers, but this benefits mainly the larger producers, who sell considerable inventory through those channels (especially large-format and bag-in-box brands). Small to mid-size wineries, on the other hand, have less of this form of retail distribution, relying mostly on bottle sales through restaurants and bars, as well as direct-to-customer business through their tasting rooms. With on-trade locations closed for a long period, revenue has been hard to come by (Despite high sales, Covid-19 has taken a toll on California's independent wineries).

So, size matters, in terms of weathering the ups and downs of the retail world. It is not just pandemics that we are talking about here — yearly variation in sales volume is “situation normal” for primary producers. There needs to be some reliable mechanism in place to address the economic hurdles created by this annual variability; and bigger companies find it much easier to jump over the hurdles. But how big is “big enough”, in the case of wineries?

Obviously, I cannot answer that question in any global sense. So, I will restrict myself to one country, as an example. That country will be New Zealand, mainly because some relevant data are available.

The Deloitte organization is a multinational auditing and consulting network, which (among other things) releases a whole range of reports about financial trends, in many places on the globe. One of those reports is based on an annual Wine Industry Benchmarking and Insights survey in New Zealand, produced in conjunction with the ANZ Bank and the New Zealand Winegrowers association.

Each year since 2006, a survey questionnaire has been sent to all members of New Zealand Winegrowers, asking about details of their previous year’s financial statement. The latest report claims that the current respondents “account for approximately 44% of the New Zealand wine industry by litres of wine produced and 35% by export sales revenue ... [they] either own or lease 39% of the 37,969 producing hectares currently under vine”. This is not a great response rate, but it is better than we might expect.

The report compiles the response data on financial aspects like supply and demand, revenue streams, profitability, equity, and return on assets, as well as other important things like key markets and customer connections. Here, I am interested solely in profitability, and its relation to winery size.

The reports subdivide the wineries into four size tiers, based on total annual revenue in $NZ: $0-$1.5 m, $1.5-$5 m, $5-$10 m, $10-$20 m, and $20 m+. [Note: $US 1 ≈ $NZ 1.5] Aggregate data are presented for each of these tiers, as shown in the graph above. This illustrates the mean profit before tax as a percentage of total sales, for each size tier (referred to as a turnover band), across all of the published surveys to date.

The basic answer to our question is, thus, that wineries need to be in the top three tiers in order to be consistently profitable, through the years. The lower two size tiers are much more risky, from year to year.

So, lets look at these lower two in more detail. To do this, we should look at the variation in the data among the wineries within each tier. In finance, this is best done by looking at the median value plus the inter-quartile range. [If this is unclear, see the explanation at the bottom of this post.] In the following graphs, each vertical bar represents one year (ie. one survey), with the middle cross-bar representing the median and the upper and lower bars representing the quartiles.

Here is the appropriate graph for the second-smallest winery size. It shows that at least 75% of the wineries made a pre-tax profit in almost every year. On the other hand, 75% of these wineries made less than 20% pre-tax profit on sales. This is not great, but it is probably sustainable, long term.

Sadly, the same cannot be said for the group of smallest wineries, as shown in the next graph. Here, in only some years did at least 75% of the wineries make a pre-tax profit (eg. 2018); and in only 6 / 13 years did at least 50% of them make a profit. Even worse, in some years almost nobody made a profit (eg. 2012); and even when a profit was made (half of the time) at least 75% of these wineries made less than 10% pre-tax profit on sales. The year 2009 was obviously very bad for at least 50% of the wineries (with 25% losing more than 250%).

So, the detailed answer to my question is that a turnover of at least $NZ 5 million seems to be good enough for long-term winery profitability, while those with less than $NZ 1.5 million will have serious trouble balancing the good years with the bad ones. You can translate this into your own favorite currency, for comparison with other countries.

Note that I have discussed only economics here, and not any of the other issues that might be related to winery size, such as regulatory compliance (eg. see the Napa Valley Save the Family Farms).

So, why do small wineries even exist, at all? Bob Campbell has noted (Not for the faint-hearted):

I recall enjoying lunch with two ex-winery owners. One had lost millions of dollars while the other confessed that he had only made a profit in one of the nineteen years he had been in business. Both were astute businessmen who knew all about budgets, cash flow and the importance of staying in the black, but both had been dazzled by their passion for wine.

---

For the data summaries, 50% of the survey responses are greater than the median, and 50% are less than the median (as represented by the dot in this figure). Furthermore, 50% of the responses are between the upper and lower quartiles (as represented by the grey box).

The study of grape-vine leaves is harder than you might think

A couple of weeks ago a new research manuscript appeared online:

Daniel H. Chitwood (2020) The shapes of wine and table grape leaves: an ampelometric study inspired by the methods of Pierre Galet.

This was recently highlighted on the American Association of Wine Economists’ Facebook page, which drew it to my attention.

I found this paper fascinating, because a whole swag of fancy data-analysis techniques were combined, in order to do something very challenging — study the variation in leaf shape among commercial grape varieties. This is of practical as well as theoretical importance, because it is this variation that has traditionally been used to identify the varieties.

I won’t bore you with the details, which really do require some expertise to understand. However, one thing did stand out to me as a bit of a worry. Figure 2 of the paper presents the results of an analysis that forces the data into a particular type of pattern, and this is not necessarily a good thing to do. There is a better alternative.

Having described the leaves of 60 grape varieties, in the manner illustrated by the red lines shown above, we can summarize the complex data by calculating a measure of the morphological “distance” between the varieties — a greater distance indicates less similarity between the varieties. This is what we want, because it is the distance that will help us study the leaf variation.

However, the analysis chosen in the paper to study these distances was what is called a cluster analysis, which aggregates the leaves into groups or clusters. This is risky, because we do not actually know that the leaves will fall into groups, in the first place. What if they don’t? We then have a result that is misleading (i.e. groups that do not exist).

We can examine this possibility by using a network analysis, instead, as described in my post Summarizing multi-dimensional wine data as graphs, Part 2: networks. This analysis makes no prior assumption about the existence of groups — if they exist then the analysis will find them, but if they don’t then it will show how complex are the non-group relationships.

My own network analysis of the distances, as provided in the paper, is shown in this graph.

The original cluster analysis found two main groups (I and II), with four outlying varieties that were in neither group. However, the network does not show us any clear groups, at all.

This does not mean that the network analysis finds fault with the cluster analysis, but merely that the cluster results are too simple — the grape varieties cannot be grouped so neatly.

I have marked the Group I varieties in red, and the Group II varieties in blue, with the outliers in black. These two groups are actually well represented in the network, as they aggregate at opposite ends of the graph. So, the cluster results are not surprising. If we are going to put the 60 varieties into two groups, then the two groups found by the cluster analysis are as good as any. The main fault, however, of the cluster analysis is that the relationships between the grape-vine leaves are more complex than this — there are any actually number of ways of clustering the varieties into groups, and the cluster analysis simply chooses one of the many.

The most obvious example of this problem is the Grenache variety, which the network analysis associates with Group I, not Group II. In the cluster analysis, Grenache is shown as the outlier in Group II, indicating that this analysis is equivocal. Unfortunately, the results of a cluster analysis cannot indicate equivocation. A network analysis, on the other hand, is specifically designed to show equivocation, if this is needed.

Equally interesting is that the network analysis shows that the four outlying varieties have, in fact, very little relationship to each other — that is, they do not appear together in the network. Burger and Chasselas do appear near each other in the graph, but the latter has a very long edge, indicating that its leaves are the most unusual within the collection of 60. Gewürtztraminer seems to be rather similar to White Riesling, while Zinfandel is associated with Gamay and Müller-Thurgau.

Conclusion about leaf shapes?

Leaf shapes have been an essential component of previous methods for the identifications of grape varieties, but they seem to be a bit more like an art than a science. This is why the modern world now uses DNA sequencing for that same purpose.

Reporting of wine-industry reports: accuracy or exaggeration?

As a scientist, I am often intrigued by commentary in the popular press concerning the contents of formal scientific reports. It is clear that several things can happen: the reporter gets it roughly right, which has happened recently for much of the Covid-19 reporting; the reporter has not really understood things, and distorts the information, often badly; or the reporter simply exaggerates the whole thing, presumably for some sensationalist purpose of their own.

It is not really any different in the wine industry. There are a number of organizations conducting industry surveys, and providing what are usually called "insights" in a subsequent report. These reports are then summarized by the online wine media, sometimes accurately and sometimes not.

Having mentioned Covid-19, I will note that Wine Intelligence has recently started releasing a set of reports that they call the Covid-19 Impact Series. These are based on formal customer surveys in a range of important wine markets, both before and after the start of the pandemic, to assess the impact of changing customer behavior. In April and May they released reports on Australia, the USA, the UK, Portugal, France, and China; upcoming reports will be on Japan (June), and the Netherlands, and Germany (July). The publishers note that:

The first three Covid-19 Impact Reports from Wine Intelligence have focused on the English-speaking markets of Australia, the US, and the UK. Comparing the data shows some remarkable consistencies, but some equally interesting contrasts.

Both the report for Australia and the one for the USA have generated some commentary in the general wine media. The commentary conclusions seem to vary from quite accurate to wildly exaggerated.

In the first category we might include this quite extensive headline from Wine Industry Network Advisor: New Wine Intelligence COVID-19 Impact Report reveals Australian wine consumption is holding up but spend per bottle is down with many wine drinkers being cautious about going out once lockdown is over. You don't even need to read the rest of the article!

In the second category we might include their equally extensive headline regarding the USA report: US drinkers have increased wine consumption during lockdown, led by more involved drinkers, as interest in locally produced wine surges. The last word in that heading is based on the following graph, illustrating one survey response. I am sure that the wineries are happy with the small increase in interest, but it is a long way from a surge. Mind you, it is better than the results for foreign wines, which all showed a decrease.

Even more interesting is this headline from 1WineDude: We’re a bunch of lushes. The author, Joe Roberts, might be this, but I am not sure about the rest of us. In defense of his opinion, he notes these reported survey results:

Now, 9.3 wine occasions per month equates to having a glass (or two) of wine with your dinner every 3.3 days. Increasing the occasions to 9.7 per month equates to once every 3.1 days, which is a 4% increase in frequency. Such people are not lushes. Myself, I am likely to have wine with dinner twice per week, which is obviously once every 3.5 days; but I am an old man, and cannot behave like young people. When I was a university undergraduate student, it could easily have been five times per week; but the quality of the wines was much worse, of course.

Of much more interest to the US wine industry are some of the other survey results in the Wine Intelligence report. I will not exaggerate them, but simply reproduce them. In each case, click on the image to see it properly.

First, the good news: wine sales are up, (but not as much as bottled water).

Now, the bad news: social drinking occasions as not going to re-appear too soon.

Finally, the even worse news: boosting savings is now top priority, not spending money.

So, no-one is expecting the US wine industry to recover from the Covid-19 pandemic anytime soon.