Thursday 13 June 2019

What counts as 'Moderate' reliability for Cochrane?

Glenn Boyle prompted me to hit the Cochrane Review's findings around SSB price interventions a bit more closely.

If this counts as moderately reliable evidence, well, draw your own conclusions about the weight to put on the stuff they count as low or very low reliability.

Here are the three studies they evaluated.

  1. Cornelsen 2017. They looked at a 10 pence price hike on SSBs across a chain of UK restaurants. 9% reduction in SSB items sold per customer. While there was a 22% increase in per-customer sales of fruit juice on the main menu, there was a 10% drop in fruit juice sales on the kids' menu, a 7% drop in sales of diet cola, and a 6% drop in sales of bottled water. They did not run it as a diff-in-diff, just looked at pre/post. Tap water isn't measured; could have been generalised shift towards tap water. And there's weird stuff - the change in consumption in levied SSBs is reported as significant with a p-value of 0.004, but the 95% CI runs from a -15.21 drop to a 3.15 increase. Typo? Who knows. It's public health. No discussion of whether there just might have been a coincidental generalised shift to tap-water given the drops in sales of diet colas that didn't get the price hike. Nothing on whether there was substitution into desserts. And it's all at Jaime Oliver restaurants; their shifting towards bankruptcy over time plus his generalised increasing awfulness may have affected customer cohort too. If his big menu reorganisation and self-imposed sugar tax was accompanied by a giant "I'm so great look what I did" publicity campaign, as it almost certainly was because it's Jamie Oliver we're talking about here, well, you might think that the clientele might have shifted away from the people who like soda to the kind of people who like folks that yell at people who like soda. No accounting for any of that. 
  2. Blake 2018. A price intervention in a single convenience store reduced sales at that single convenience store. It was the convenience store in a hospital, so maybe it was a bit harder for folks to pick other stores. But come on. 
  3. Breeze 2018. Leisure centres in Sheffield increased SSB prices by 20 pence and saw a 31% reduction in sales per customer, with some increase in sales of diet soda. No clue how many people brought in their own soda in their gym bags. Oh, but the intervention also included staff training and publicity and stuff that might have had additional effect on their own around salience of health and price.

But when sugary drinks are made more expensive, or sugar-free alternatives made cheaper, sales fall, the researchers found.

"The evidence is unequivocal... you put up the price, consumption goes down," NZ Dental Association spokesperson Rob Beaglehole told The AM Show on Thursday.

"You get rid of junk food from schools, consumption goes down. Better sugar labelling is again another way of reducing sugary drink consumption. There's lots of different ways that we can act."
I'll never disagree that demand curves slope down. But the elasticity matters. And this latest Cochrane Review ... well, if that's the basis for recommending generalised hikes in SSB prices via soda taxes...

No comments:

Post a Comment