Thursday 13 June 2019

Sweet restrictions

The Science Media Centre asked me for comment on the latest Cochrane Review on interventions around sugar.

Reading through the thing, I was struck by the weakness of evidence around a lot of the kinds of things folks here like to demand that the government do.

Cochrane rates the certainty of evidence on a scale that runs: very low, low, moderate, high.

The interventions with the strongest evidence base around environmental interventions aimed at reducing consumption of sugar-sweetened beverages were rated "moderate". Nothing rated high. Interventions rated as having "moderate" evidence included:
  • Improved access to low-calorie beverages in the home environment
    • studies in this group would provide free home delivery of bottled water or diet drinks to people often in places with unreliable access to clean drinking water, and found reductions in soda consumption as consequence. 
  • Multi-component community campaigns focused on SSBs
    • Results here drew from one study. 
  • Government food benefit programs with incentives for buying fruit and vegetables and restrictions on the purchase of SSB
    • Here, studies looked at interventions restrictions on purchasing SSBs using the equivalent of New Zealand's WINZ payment card. The studies found reduced sugar consumption. New Zealand already bans a lot of classes of purchase on the payment card, including alcohol and tobacco. It wouldn't be infeasible to do it here, but there would be a lot more SKUs that would have to be loaded up properly into the card - it could prove difficult in practice. Checkout clerks are already well trained around identifying alcohol and tobacco purchases; knowing which beverages would be banned and which would not would require the back-end systems being programmed correctly. I expect it wouldn't be a simple thing.
  • In-store promotion of low-calorie beverages in supermarkets
    • Evidence here all drew from Foster 2014. That was a randomised trial looking at in-store promotion of healthier items: lower-fat milk, ready-to-eat cereal, frozen meals, in-aisle beverages (Diet Pepsi and Aquafina water), and checkout cooler beverages (zero calorie beverages and water). Effects were often statistically significant, but I'm not sure that a store selling 24 more gallons of skim milk and 53 more gallons of 1% per week are really all that big a deal. 
  • Price increases on SSB
    • Three studies found that places chosen for soda price interventions, like a leisure centre or a particular corner store, saw reduced sales of those price-boosted items. But it seems kinda likely that folks would just be purchasing their soda at shops next door that weren't part of an experiment involving higher soda prices. You oughtn't generalise from it. No surprise that Beaglehole does generalise from it though
  • Small prizes for the selection of healthier beverages in school cafeterias
    • Evidence ranged in strength from moderate to low. In the Hendy 2011 study (rated moderate; the others were low), there was a three-meal-per-week reduction in the number of meals with unhealthy beverages selected in an intervention in primary schools where kids' meals were monitored and they got token rewards from parent volunteers chosen as monitors. If your school has that many available parent volunteers, I guess it's not that high cost to implement - though you might imagine other things parent volunteers in schools might more usefully help with. 
  • Traffic light labelling
    • Evidence here was rated moderate in reducing consumption of red-labelled beverages. 
Everything else had evidence rated as low or very low confidence. 

Here's what I told the Science Media Centre about it; Newshub picked up a bit of this commentary but didn't contrast the price bit I'd noted with how Beaglehole approached it.
The Cochrane Review provides an important synthesis of the evidence regarding non-tax interventions aimed at reducing consumption of sugar-sweetened beverages [SSBs].

The review found that many often-recommended measures have little evidentiary base, with certainty of evidence rated as very low. Interventions in this category included measures like healthier vending machines in workplaces and schools, restrictions on the number of stores selling SSBs, urban planning restrictions on new fast-food outlets, and menu-board calorie labelling. No studies were found that might provide basis for restrictions on advertising.

Some measures showed promise, with a moderate certainty of evidence established across numerous studies.

Improved access to low-calorie beverages in the home environment reduced SSB consumption, but many included studies focused on places without reliable access to clean drinking water. Regular home delivery of free non-SSB drinks across broad swathes of the population seems unlikely to pass any reasonable cost-benefit assessment, and especially in places where piped water is of reasonable quality.

Restrictions placed on purchases funded through food benefit programmes reduced SSB consumption, and could be implemented in New Zealand by adding SSBs to the list of prohibited purchases on Work and Income Payment Cards. But the administrative costs may not be trivial, and the imposition on low-income households who enjoy soda occasionally should not be ignored.

Small prizes for selecting healthier beverages in primary school cafeterias showed some promise.

While price increases in individual targeted stores showed reduced sales of SSBs in those particular venues, the surveyed studies in that area do not look at overall consumption; people could easily have shifted to purchasing from outlets where prices had not been hiked.

And while the review authors tentatively suggested a somewhat broader set of interventions may prove effective, they also warned that their confidence in the likely effects is low to moderate. Rather than providing evidence for policy change, we should view the report as suggesting measures potentially worth trialling within an appropriate experimental framework designed to improve the evidence base.

Where the evidence base presented for any substantial effect of interventions is moderate at best, even without being evaluated as part of a broader cost-benefit assessment that weighs implementation costs and costs to consumers, we should be highly sceptical of any calls for strong intervention based on this report. It should rather temper our enthusiasm for large-scale measures likely to impose substantial cost for rather less certain benefit. Pilot studies and trials of some of the more promising interventions may be warranted.

No comments:

Post a Comment