Latest study underplays the role of the curfew and movement restrictions in the decline of unnatural deaths during Lockdown

Media Statement

The alcohol industry notes a recent study (“the Moultrie paper”) by the Centre for Actuarial Research and Medical Research Council (MRC), published in the South African Medical Journal, that concluded that alcohol bans, and not the curfew, caused an observed decline in unnatural deaths – such as motor accidents and violent murders ‒ during the lockdown. The study attempts to counter earlier analysis, funded by the alcohol industry, which looked at the relationship between trauma admissions, alcohol restrictions and curfews (including mobility) during varied adjusted COVID restrictions using regression analysis.

The industry’s own in-depth analysis, which was conducted by independent analysts and peer reviewed, indicated that mobility tracked hospital admissions, and trauma admissions likely dropped because people are at home and under curfew rather than being out and about. In addition, the industry analysis also considered the counter-factual that several countries that did not impose alcohol bans as part of their response to the COVID-19 pandemic, achieved the same level of reduction in trauma admissions as South Africa (60%). These included the UK (57%), Italy (57%), and Ireland (62%).

The industry has argued that restricting mobility through curfews, curbing gatherings and other protective measures including hygiene and sanitation have been recognised globally as the most effective and pragmatic approach to halting the spread of COVID-19.

The alcohol industry has legitimate concerns about the Moultrie paper, including the data it relied upon.

Following a statistical review of the paper, a concern was expressed that the raw data” used in the study is in fact data that has been calculated using a model constructed by the MRC itself (several co-authors of the Moultrie paper are party to the estimation of the “raw-data”). This raises questions about any conclusions drawn from any subsequent analysis based on this data, as the data has been generated by the MRC in the first place.

The study ignores restrictions implemented alongside alcohol bans

Crucially, the focus of the Moultrie paper is to try and show that it is alcohol restrictions and NOT issues of mobility/curfew that drive the reduction in unnatural deaths. It is clear that, BOTH of these factors will have some impact on unnatural deaths ‒ the only argument could be, which is the predominant cause? But, because in almost all cases both mobility restrictions AND alcohol restrictions are imposed together, it is statistically problematic to distinguish between these two effects. Statisticians say the two variables are “confounded” in their impact on unnatural deaths.

The Moultrie paper seems to ignore the impact of simultaneous restrictive curfews over the same period as alcohol restrictions. Furthermore, the analysis seems to be dependent on specific curfew categorisations. Clearly, slicing the data into different curfew buckets will lead to different results.

All four total bans of alcohol sales (including the current ban which has a 21h00 to 04h00 curfew) in South Africa have been implemented with a corresponding curfew of seven hours or more. The study, however, introduces curfew scenarios of less than seven hours—which did not exist in relation to the bans on alcohol sales in South Africa—and, therefore, the study systematically discards the role of the curfew in the reduced number of unnatural deaths that it observes and incorrectly concludes that “the sale of alcohol rather than curfew is associated with the reduction in unnatural deaths observed during the Covid-19-outbreak in SA”.

Of particular note is that the study found a further 13% decline in unnatural deaths when the country moved to Level 3 and the alcohol ban was lifted. During this period, alcohol was sold for home consumption from Mondays to Thursdays in all the bottle stores and e-commerce channels. Taverns, which largely have on-site consumption licences, were allowed to sell for home consumption only, in order to allow access to alcohol sales in townships and rural areas.

Significantly, the study fails to explain what factors drove this decline in unnatural deaths while alcohol was being sold.

Cause of deaths not recorded

The methodology used in the study is also problematic. The authors of the study admit that there is no information about the actual cause of death. It is therefore impossible to determine whether alcohol was at all involved in the deaths recorded or the proportion that were alcohol-attributable.

Alcohol restrictions do have a likely influence on the number of drunk-driving accidents, for example, but it is also true that these incidents declined because of curfews. To attribute this to the alcohol ban alone is therefore misleading.

No reference to global experience

We note that the researchers avoided drawing on international literature available and drew no comparisons with what they were investigating in South Africa. This is despite the fact that COVID-19 is a global pandemic and with a number of containment approaches applied in different regions.

Of even greater concern, in the study conclusion, the researchers advocate for policy changes such as restricting alcohol advertising and increasing the current legal drinking age of 18 years. The study on whether the sales bans or the curfew had greater bearing in the reduction in unnatural deaths has no relevance to whether drinking should be allowed at the age of 18 or 21 years, which is a separate research question and does not play a role in determining policy around a pandemic response.

The alcohol industry calls on the government to reassess its response to the coronavirus pandemic by implementing more effective measures, including addressing its handling of the vaccine rollout, that will have greater success in curbing the sharp rise in infection rates during the third wave.

Note: This release was made using statistical commentary from Professor Graham Barr, Emeritus Professor in the department of Statistical Sciences at University of Cape Town, who has reviewed the Moultrie paper.

Issued by FTI Consulting on behalf of the South African Liquor Brandowners Association (SALBA), which represents various alcohol manufacturers including Distell, Heineken, Diageo, Pernod Ricard.

salba@fticonsulting.com