In the first weeks of South Africa's 2020 alcohol sales ban, Johannesburg's forensic mortuary took in 54% fewer unnatural deaths than in the same weeks a year earlier. These were the crashes, the shootings, and the assaults.

I don't usually write about policy. Bans aren't my thing, and I'm building an app on the opposite premise. But this paper is hard to ignore. It shows what alcohol is tied to across a whole city, and the answer has nothing to do with livers or long-term risk. It's about who ends up on a forensic pathologist's table on a Saturday night.

What they did

During COVID-19, South Africa banned alcohol sales four separate times. Each ban started and stopped on a set date. That left the country with an unplanned experiment: the same city, with and without legal alcohol.

Researchers pulled every unnatural death investigated by the Johannesburg Forensic Pathology Services Medico-Legal Laboratory during those four ban periods. "Unnatural" here means road crashes, pedestrian deaths, gunshots, blunt trauma and the like. Deaths from illness weren't counted.

Then they matched each ban period with the same calendar window a year before, when sales were normal, and broke the counts down by sex, population group, age, circumstances of death and medical cause.

This is an ecological study. It compares totals across periods. It doesn't track individual people or their blood alcohol levels.

What they found

  • First ban: unnatural deaths dropped 54.2% compared with the same period in 2019.
  • All four bans together: a 26.4% drop overall, with fewer deaths in every demographic group.
  • Who: the significant drops were among men, Black and Coloured South Africans, and people aged 21 to 40.
  • How: deaths from motor vehicle crashes, pedestrians hit by vehicles and firearm discharges all fell significantly.
  • Cause: blunt force trauma and gunshot wounds too.

That demographic list is basically who makes up most nights out in any big city. Young adult men.

What it means

The big caveat comes first, because it's a real one. The first ban landed alongside the hardest lockdown South Africa had. People couldn't go out. Roads were empty. Police were everywhere. So part of that 54% belongs to the lockdown, and this design can't tell you how much. The later bans came with looser restrictions and the effect shrank to 26.4% overall, which fits that reading. (The data also come from one city, and the abstract doesn't mention adjusting for population changes between the years.)

Still, I don't think the signal is noise. Deaths fell in every group. And the drops sat in exactly the categories you'd tie to drinking: crashes, people walking home and getting hit, guns going off, fists and blunt objects. It lines up with an umbrella review I wrote about earlier, which found that any drinking raises the odds of injury about 2.8 times. Johannesburg is that number playing out across a whole city.

I'm not reading this as a case for a ban. A ban is a blunt tool. It cuts supply for everyone, including the many people who drink and never hurt anybody. The harm in this data comes from specific nights, not from one drink with dinner: someone far past their peak, deciding about a car, a walk home, or an argument.

That's also where personal judgment fails hardest. A recent RCT (randomized controlled trial) found that the most intoxicated people were the worst at estimating how drunk they were. You need an honest read on yourself most at the exact moment you're least able to give one.

So there's another lever, and a government can't pull it for you. AlcoBalance is built around it. On a long night the pace creeps up without you noticing. The app shows where you are on the curve and how long until your level fades. You decide on the next drink, or on how you're getting home, with that picture in front of you. Beats guessing from inside your own blind spot.

A city-wide ban showed how many deaths cluster around those few hours. I'd rather people could see those hours clearly than have the shops close.

Source: International Journal of Injury Control and Safety Promotion, DOI