STATS ARTICLES 2012
Will Minimum Pricing Reduce Alcohol Deaths?
Rebecca Goldin, Ph.D., February 28, 2013
Public health advocates are hailing a new Canadian study that claims minimum pricing could slash deaths from alcohol. But a close look at their statistical methods and data sourcing raise more questions than answers
Would raising the price of alcohol by setting a minimum price reduce alcohol related death? Off the top of your head, you might think so: cost goes up, demand goes down – unless, of course, the kind of people who are the most likely to die from abusing alcohol are highly resistant to changes in price, in which case all the government does is collect more money, and possibly deter moderate drinkers.
But the world of alcohol policy was recently shaken by a new Canadian study that seemed to nail the data. As Reuters reported, “Increasing the minimum price of alcohol by 10 percent can lead to immediate and significant drops in drink-related deaths and may also have long-term beneficial health effects.”
And one of the study authors, Dr. Tim Stockwell, who is director of the University of Victoria's Centre for Addictions Research of British Columbia, told the BBC: “This study adds to the scientific evidence that, despite popular opinion to the contrary, even the heaviest drinkers reduce their consumption when minimum alcohol prices increase. It is hard otherwise to explain the significant changes in alcohol-related deaths observed in British Columbia.”
Yet the observation that heavy drinkers are less sensitive to price than moderate or light drinkers has been confirmed through a variety of means – so the Canadian evidence would really have to be strong to call this conclusion into question.
So, let’s look at this new evidence. According to the study – The Relationship between Minimum Alcohol Prices, Outlet Densities and Alcohol Attributable Deaths in British Columbia, 2002 to 2009 (Zhao et al.) – “a 10% increase in average minimum price for all alcoholic beverages was associated with a 31.72% (95% CI: 25.73%, P<.05) reduction in wholly AA deaths.” On the face of it, that is an impressive effect. But a closer read raises questions about how the authors parsed the data, and whether the observed reductions can really be attributed to alcohol price increases at all.
As a matter of statistics, deaths in British Columbia stayed constant during the period in which prices increased, and for the most part, so did prices – alcoholic beverages as a whole roughly tracked with the Consumer Price Index. This can be seen in the data put out by Statistics Canada.
And yet the authors of the study were able to measure a correlation between increased prices and decreased death over the time period of 2002-2009. How is this possible? It’s all in how you disaggregate the data.
Disaggregating the data
The study disaggregated the data by region, by type of alcohol-related death, and by the type of alcohol. Between 2002 and 2009, the minimum cost of spirits went up by 18 percent, and beer by just under 16 percent, after adjustment for inflation, according to data they obtained from the Liquor Distribution Branch (LDB) of the BC Ministry of Public Safety and Solicitor General. As they say themselves, the average prices do not seem to correspond well to overall alcohol consumption. So while average prices tracked with inflation, apparently minimal prices for beer and spirits went up rather substantially.
They authors claim that there is an observed decrease in “wholly alcohol attributable (AA)” deaths that tracks closely with increased prices of beer and spirits. They did not find the same effect with either “acute” alcohol deaths or “chronic” alcohol deaths, leaving open the question of how robust the results are. Wholly AA deaths include death by alcoholic psychoses, alcohol dependence, alcohol abuse, alcoholic cardiomyopathy, alcoholic gastritis, chronic pancreatitis (alcohol induced), fetal alcohol syndrome, excess alcohol blood level, accidental poisoning and exposure to alcohol, and intentional self-poisoning by and exposure to alcohol. Some of these deaths are “acutely” caused by alcohol while others are chronically caused by alcohol. Though other researchers seem to indicate that “wholly” AA deaths include death by various malfunctions of the liver, Zhao et al. don’t include liver complications.
The overall picture looks very different. According to the British Columbia Vital Statistics Agency, the number of all alcohol-related deaths between 2002 and 2008 remained steady, at roughly 5.4-5.6 deaths per 10,000 people. Then there is a drop in 2009 to 4.7 deaths per 10,000 people. All, in this case, means direct causes (alcohol intoxication, alcoholic psychoses and dependence, alcoholic cardiomyopathy, alcoholic gastritis, alcoholic liver disease, alcohol induced chronic pancreatitis, and alcohol poisoning) and indirect, (deaths by certain infectious and parasitic diseases, neoplasms, endocrine/metabolic, mental disorders, neurological diseases, circulatory, disease of the respiratory system, digestive system diseases, urinary system diseases, unintentional injury, suicide, homicide or “other”).
In a previous article, several of the same authors as the Zhao et al. study claimed that alcohol consumption was going up – in fact, the price increases that Zhao points to in 2004 and 2006 seem to have happened at the same time that alcohol consumption actually increased. This may not be the case for spirits and beer – the two types of alcohol that saw an increased minimal price – but the overall picture shows people drinking more, not less.
In this newer study, Zhao et al. parsed the data in several different ways: they looked at 16 regions within British Columbia, rather than examining the data all together. But the authors report only the aggregate numbers: 3,642 acute deaths and 5,842 chronic deaths, and between these two groups 1,388 “wholly” alcohol attributable deaths. Because wholly alcohol attributable deaths include some acute deaths and some chronic deaths, the numbers in Zhao’s studies are different than those obtained from the standard statistical outlets. This makes it difficult to account for why there is a correlation between price increases and “wholly AA death” decreases, but why such a correlation doesn’t exist for other categories of death.
When one looks at acute alcohol deaths or chronic deaths, there is no effect from price increases. In particular, the statement that a price increase in 10 percent would reduce death by over 30 percent seems highly suspicious considering that the prices of liquor and beer went up far more than 10 percent, and yet deaths overall didn’t go down between 2002 and 2008 – and only saw a drop from 2008 to 2009.
Multiple Testing and Lagging Impact
The study also claims that the impact of price changes can be seen over time. The authors provide a table (Table 4 in Zhao et al.) showing a highly significant association between increased price and decreased death by wholly attributable causes in the 4th quarter following each price increase.
What’s troubling here is that they break the data down into many quarters and categories, run multiple statistical tests, but don’t adjust for multiple testing. This results in a table spotted with statistically significant results even as basic statistics tells us this method will produce spurious results.
A close look at the table is suggestive that spurious results are indeed at hand. This table looks at 16 quarters following a minimal price increase, and whether there is a correlated increase or decrease in deaths among acute, chronic or wholly attributable alcohol deaths. The authors point to a statistically significant decrease in wholly attributable deaths in the quarter that a price increase was implemented, as well as in the second and third subsequent quarters. (but not in the first quarter, nor the 4th-15th quarters).
But it also shows a significant increase in acute deaths in the third and fifth quarters after a price increase, and then a statistically significant decrease in acute deaths in the 8th quarter after the price increase. Chronic deaths saw statistically significant decreases in the 8th, 9th and 13th quarter after a price increase, but not in the other 13 quarters. This all suggests that the results could be, at least in part, the result of simply running a lot of tests on a lot of data – and without adjusting for multiple tests, randomness can creep in. Though the results do lean toward decreased death, picking out the most extreme of the results (as the media and the authors of this study did) may be misleading. In fact, if the 1,388 wholly alcohol attributable deaths occurred evenly over the quarters, these numbers refer to trends in about 87 deaths each quarter – trends that would be highly sensitive to a small number of deaths.
So while there seems to be an overall trend of decreased death with increased prices, the failure to account for multiple testing means there could be true correlation or there could be just a statistical fluke.
Demographic Shifts During Measurement
British Columbia’s minimal pricing did not have an impact on all products. Only spirits and beer saw price increases, whereas wine and coolers did not. At the same time, privatization of state liquor stores meant that there was an increased density of stores in certain areas, which the authors associated with a 2 percent increase in acute and chronic deaths. Is this finding reliable? Unfortunately, the authors fail to tease out the impact of this change in alcohol availability from the impact of rising prices in a convincing way.
Demographic changes can have confounding effects that are not accounted for by Zhao et al. As an exercise in how statistics can go wrong, let’s suppose that no one is price sensitive. However, those who are alcohol lovers tend to want to move into areas with increased density of liquor stores following privatization (or the density increases in areas where the heaviest alcohol consumers live). Then one might see an increase in death in these areas with more liquor stores, not because the existence of the stores led to more drinking, but rather because the drinkers became more localized. Furthermore, the folk left behind drink less – which makes regions with fewer outlets look like they’ve been impacted by the price increase even more. In other words, in this hypothetical scenario no one was in the least changing their drinking patterns due to price changes, yet the data show decreased deaths occurring with price increases.
This Canadian study was greeted as a breakthrough by public health lobbyists, particularly in the UK, where politicians are debating imposing minimum alcohol prices. Unfortunately, the study was filled with multiple testing and weak explanations for the whole data set, rather than small subsets. The study failed to provide the data and information necessary to validate its conclusions, leading us at best to say that the results might be true or they might be spurious. Rather than swallowing whole a tale by which market forces can greatly influence deaths from drinking, the media ought to have looked more closely: they would have found more questions than answers about alcohol deaths and its relationship to alcohol prices.
Rebecca Goldin, Ph.D. is the Director of Research at STATS.org. Dr. Goldin was supported in part by National Science Foundation Grant #202726