Hide table of contents

Note: This report has been superseded by our subsequent Deep Report on Diabetes - please browse that report for our most updated findings.


Taking into account the expected benefits of eliminating diabetes mellitus type 2 (i.e. improved health and increased economic output from eliminating DMT2 itself, as well as improved health and increased economic output from reduced depression for which DMT2 is a risk factor), as well as the tractability of policy advocacy for a sugar-sweetened beverages tax as well as other World Health Organization (WHO) sugar reduction Best Buy interventions (i.e. mandatory nutrition labelling and a mass media campaign to reduce sugar consumption), CEARCH finds that the marginal expected value of policy advocacy for these top sugar control solutions to eliminate DMT2 to be 11,036 DALYs per USD 100,000, which is around 20x as cost-effective as giving to a GiveWell top charity (CEA).

The full 35-page report may be found on CEARCH's website (report); this post is a high-level summary intended for forum readers.


Key Points

  • Importance: This is a strongly important cause, with 1.47 * 1010 DALYs at stake from now to the indefinite future. Around 98% of the burden is directly related to diabetes mellitus type 2 (DMT2) itself (inclusive of both the health and economic burden), while 2% relates to the downstream consequences of an increased risk for depression (again, inclusive of both the health and economic burden)
  • Neglectedness: The present global disease burden of DMT2 is expected to grow as a result of various factors – especially economic growth (which causes a shift to processed foods, sedentary lifestyles, and greater obesity rates), ageing (since the pancreas, which produces insulin through the beta cells, is exhausted with age) and population growth. At the same time, governments are doing little to solve the problem, with existing government policies around the world with respect to high-impact sugar-sweetened beverages (SSB) taxes expected to solve perhaps 0.4% of the future global DMT2 burden. Going forward, governments are counterfactually passing more of such SSB taxes at around only 2% per annum, which is not sufficiently rapid as to combat the growing disease burden.
  • Tractability: A solution with moderate tractability is available, in the form of policy advocacy for SSB taxes and the WHO's Best Buy interventions for reducing sugar consumption. While the chances of policy advocacy for such SSB and sugar consumption reduction policies is fairly low, the empirical literature strongly supports the idea that such interventions will help reduce SSB and sugar consumption and that lower SSB and sugar consumption will in turn reduce the disease burden of hypertension.

Cluster View

  • Beyond the headline CEA results, it can be valuable to use cluster thinking (i.e. looking at the issue from multiple perspectives) to inform our decision on whether to channel resources towards a cause area – as CEAs are sensitive to errors, and multiple perspectives offer a way to robustly sense check our quantitative estimate. Employing this cluster view, and looking at the quality of evidence, expert opinion, and multiple critical considerations, CEARCH finds that DMT2 as a cause area does indeed look extremely promising.
  • Quality of Evidence: Overall, the evidence suggests that the intervention is an effective one. For more details, refer to the Tractability section in the report; but here is a brief summary:
    • Success rate of SSB & sugar consumption reduction policy advocacy: Using both an outside view (e.g. of past SSB and sodium tax advocacy attempts and of general lobbying attempts) as well as an inside view, our best guess is that policy advocacy for SSB & sugar consumption reduction has a 7% chance of success.
    • How SSB and sugar consumption reduction policies affect SSB and sugar consumption: Based on various meta-analysis and systematic reviews, and after robust discounts and checks (e.g. for a conservative theoretical prior of a null hypothesis; for publication bias; and for miscellaneous issues like endogeneity), we expect: (a) a tax rise that increases SSB retail prices by 20% to reduce the consumption of SSBs by around 17%; (b) mandatory nutrition labelling (directed at reducing sugar consumption) to actually reduce the consumption of sugar by 6.2 grams/person/day; and (c) a mass media campaign (directed at reducing sugar consumption) to actually reduce sugar consumption by 1.1 grams/person/day. Note that we are least confident of (c), where a significant degree of extrapolation is required, and have discounted the estimate accordingly.
    • How SSB and sugar consumption affects health: Similarly, various meta-analysis and econometric analysis – even after the same sort of robust discounts and checks employed above – suggest that (a) reducing consumption of SSBs reduces the global disease burden of DMT2, with 4.5% of diabetes mellitus type 2 DALYs attributable to SSB consumption; and that (b) reducing sugar consumption reduces the global disease burden of DMT2, with every 150 kcal/person/day increase in sugar availability (equivalent to 38.76 grams of sugar/person/day) associated with increased diabetes prevalence by 1.1%. Note that we are only moderately confident of (b) given the evidence we relied upon being a single study conducting panel-data analysis (such that both being an unrepresentative outlier as well as being undermined by endogeneity, are significant concerns), and we have discounted accordingly.
  • Expert Opinion: The experts we consulted agreed that the global disease burden of DMT2 is expected to grow in the coming decades, and that preventive solutions – of the sort the WHO recommends and which we assess here – are generally preferable to treatment. As one expert puts it, dealing with diabetes and obesity with drugs is like trying to deal with lung cancer via medication – the real problem is the food environment. For more details, refer to the discussion on the growth of the problem in the Expected Benefit: Improved Health from Eliminating Diabetes Mellitus Type 2 section of the report, and to the discussion on potential solutions in the Problem & Solution section.
  • Multiple Considerations: Beyond the major considerations (i.e. importance, neglectedness and tractability) – as already discussed above, and which DMT2 as a cause area scores well on – DMT2 also looks promising given various miscellaneous considerations. In particular, we find that potential lines of criticism (e.g. whether the sugar reduction policies being pushed leads to substitution with respect to home foods or salty food; or whether we pay too heavy a price with respect to diminished freedom of choice and pleasure from eating and drinking sugary food; or whether an SSB tax in particular is regressive) are not sustained. For more details, refer to the Miscellaneous Considerations section of the report.



  • The marginal expected value of DMT2 as a cause area for effective altruism has dropped drastically from 1000x GiveWell at the shallow research stage to 20x GiveWell here (i.e. a 50x drop). Here are some of the factors that caused that drop, or that – conversely – led the drop to not be as large as it would otherwise have been.
  • We model the increased risk of depression as a result of DMT2, even if the relative health and economic burden of the latter is fairly insignificant compared to the direct health and economic burden of diabetes mellitus type 2.
  • A fairly significant discount absent at the shallow research stage but introduced here is a discount for the fact that we may merely be speeding up the implementation of such policies
  • We have significantly downgraded our confidence in the success rates of policy advocacy success in this space.
  • However, given that the intervention being assessed is no longer just an SSB tax but that as well as multiple effective sugar consumption control policies – which makes sense, insofar as advocates can push more than one policy at a time, when engaging governments – our tractability figures do get a boost.
  • We now explicitly discount for the time lag between start of advocacy and policy change, and between policy change and DALY impact given the damage of past diabetes amongst existing sufferers is already baked in.
  • We also take into account the fact that SSB consumption has negative health consequences beyond increasing the risk of DMT2.
  • We have significantly increased our estimate on costing, especially since we have re-evaluated the likely amount of time needed for lobbying to succeed (or to evidently fail in such a way as to warrant a shutdown or pivot) as 3 years, rather than just 1.



  • Point estimates are sensitive, and while relying on them is reasonable given that we are ultimately interested in mean estimates, caution is also warranted.
  • On top of the initial shallow research round where one week of desktop research was conducted, this intermediate research round saw two weeks of desktop research and expert interviews carried out. We have reasonable confidence in our results here, but more research at the deep stage will be needed before we will have high confidence in these findings.
  • The headline cost-effectiveness will almost certainly fall if this cause area is subjected to deeper research: (a) this is empirically the case, from past experience; and (b) theoretically, we suffer from optimizer's curse (where causes appear better than the mean partly because they are genuinely more cost-effective, but also partly because of random error favouring them, and when deeper research fixes the latter, the estimated cost-effectiveness falls). As it happens, CEARCH intends to perform deeper research in this area, given that the headline cost-effectiveness meets our threshold of 10x that of a GiveWell top charity.


Full Report

The full 35-page report may be found on CEARCH's website (report).

Sorted by Click to highlight new comments since: Today at 7:49 AM

I was wondering if you considered anyway to take into account "adversarial dynamics" (i.e., the industry increasing its investments in lobbying against such measures) and substitution effects (e.g., people spending more on other harmful products, such as alcohol)

Hi Ramiro,

Those are good questions!

(1) For substitution effects, we looked at (a) substitution with respect to home foods (i.e. the fear is we make outside food less sweet, so people just make their own food at home and add lots of sugar or sweet sauces); and (b) Substitution with respect to salty food (which leads to hypertension etc).

(a) For substitution with respect to home foods: We found that this is likely not a material risk insofar as:

  • (i) Our taste for sweetness is adaptive, and reducing sugar intake makes high sugar food taste too sweet even as low sugar food tastes sweeter than before. This is in line with what is the case for salt, where the phenomena of desensitization also exists.
  • (ii) A mass media campaign will be looking to address this precise issue, and to the extent we expect behavioural change with respect to highly processed food, we have equal reason to expect change with respect to seasoning of home foods.

(b) For substitution with respect to salty food: The evidence with respect on the cross-price elasticity is mixed, for as Dodds et al note: "A US study found that nutrient taxes targeting sugar and fat have a similar impact on salt consumption as a dedicated salt tax, likely because many foods, especially junk foods, that are high in sugar are also high in salt."; in contrast "A New Zealand experimental study ... [finds] that salt taxation led to a 4.3% increase in the proportion of fruit and vegetables purchased, but also a 3.2% increase in total sugars as a percentage of total energy purchases." Given this, we took that it would be reasonable to assume no net benefit or cost with respect to sugar consumption.

(c) For alcohol, we didn't look at this explicitly - though per Teng et al, the evidence is mixed, and from my own sense from going through the literature is that there is a small but significant substitution effective, and this will have to be modelled more explicitly going forward.

(2) On industry - we do look at the role of industry in general, and consider it an important factor that made us downgrade our estimate of the chances of advocacy success.

Thanks for the post. I find it weird that we sort of neglect scalable interventions regarding non-communicable diseases (except for tobacco).

I was hoping that after covid-19 this would become a priority. Btw, I noticed that you do not use evidence associated with the pandemic - even though DMT was one of the main predictors of mortality:
Diabetes is most important cause for mortality in COVID-19 hospitalized patients: Systematic review and meta-analysis - PMC (nih.gov)

Diabetes prevalence and mortality in COVID-19 patients: a systematic review, meta-analysis, and meta-regression - PubMed (nih.gov)
Of course, the main datasets on the burden of DMT-2 will not include covid mortality. But have tried looking into this?

I think the short of it is that trying to model the counterfactual impact of a reduction in diabetes prevalence on reduced COVID-19 burden would be too tough, relying as it does not just on complex epidemiological modelling but also inherently unknowable future scenarios. None of the experts we talked to raised this as a live issue, in any case, so my assumption was that post 2020-2022 it's not that significant compared to the global disease burden of DMT2 itself, especially on a long term basis.

Just briefly skimmed, but I was wondering if the decreased happiness from the lesser consumption of crap was factored into the analysis.

Yep! We looked at whether we are reducing people's pleasure from eating sweet food, but the evidence suggests this shouldn't be an issue - since our taste for sweetness is adaptive, and reducing sugar intake makes high sugar food taste too sweet even as low sugar food tastes sweeter than before. The upshot is that recalibrating everyone at lower levels of sugar will leave food tasting subjectively much the same over medium-to-long term.

This is in line with what is the case for salt, where the phenomena of desensitization also exists.