Affiliate disclosure

Book titles on this page link to Amazon. As an Amazon Associate, DataField.Dev earns from qualifying purchases — at no additional cost to you.

Chapter 12 — Further Reading

Real sources only. Where I'm unsure of exact details, I describe the source rather than inventing a citation.


Start with the current guidance

Read your own country's guidance, and then read one other country's. The differences are instructive and the direction of travel is consistent.

UK Chief Medical Officers' Low Risk Drinking Guidelines (2016) — gov.uk. Notable for stating plainly that there is no completely safe level, and for the accompanying evidence review, which is readable and shows its reasoning.

Canada's Guidance on Alcohol and Health (Canadian Centre on Substance Use and Addiction, 2023) — ccsa.ca. The most substantial recent revision, and the clearest example of the shift from a threshold model to a continuum of risk. Read the technical report alongside the summary; the reasoning for the change is laid out explicitly. For: everyone. This is what "the guidelines moved" actually looks like.

Dietary Guidelines for Americans 2020–2025dietaryguidelines.gov, alcohol chapter.

WHO statement on alcohol and health (2023) — who.int. Short. Read it, then read the coverage it generated, and note the gap between "no threshold below which risk is zero" and how it was reported.


On the J-curve and its collapse

The Mendelian randomization literature. Search PubMed for "Mendelian randomization alcohol cardiovascular" and for studies using ALDH2 in East Asian cohorts. Read at least one in full, including the assumptions section — MR's limitations are stated honestly by its practitioners and understanding them is what separates using the method from citing it. For: students and clinicians. This is the evidence that changed the field.

On sick-quitter bias: search for "abstainer bias alcohol mortality" and for analyses restricting comparisons to lifelong abstainers. The papers examining how the J-curve behaves under different definitions of the reference group are the cleanest demonstration of the problem.

The 2018 Lancet Global Burden of Disease alcohol analysis — the one concluding that the level of consumption minimizing health loss is zero. Read it alongside at least two critical responses; the disagreement is about modelling choices and framing rather than about the underlying data, which makes it a good exercise in distinguishing those.

The MRC Integrative Epidemiology Unit at Bristol (bristol.ac.uk/integrative-epidemiology) publishes accessible material on Mendelian randomization generally. If §12.5 was the most interesting part of the chapter for you, start here.


On cancer

IARC monographs on alcohol consumption and ethyl carbamate (monographs.iarc.who.int). The primary source for the Group 1 classification. Free. Read the summary of evaluation rather than the whole monograph unless you have time.

World Cancer Research Fund / AICR Continuous Update Project (wcrf.org) — their alcohol and cancer material, with explicit grading of the strength of each association. Their treatment of the breast cancer dose-response is the clearest available. For: anyone weighing the cancer risk specifically, and particularly anyone with a family history.


On the halted trial

Search for reporting and institutional review documents on the NIH moderate alcohol trial (MACH15) and its termination. The New York Times reporting and the subsequent NIH advisory committee findings are both findable and both worth reading.

This is one of the better-documented examples of funding influence being caught and acted on at a major research institution — and, per Chapter 1 §1.5, it is a story about both a real problem and a working correction mechanism. For: anyone interested in how research questions get shaped before any data is collected.


On the industry's role

Marion Nestle's writing (foodpolitics.com) covers alcohol industry funding of nutrition research alongside food industry funding, and applies the same standard to both.

Search for published analyses of alcohol industry funding of health research and of industry-funded social aspects organizations. Read them noting the distinction §12.4b draws: amplification and framing rather than fabrication, which is subtler and more effective.

And apply Chapter 1's rule symmetrically: look for whether comparable funding influences operate on the abstinence-advocacy side. The honest answer is that they're smaller — there is far less money in people not drinking — but "smaller" is not "absent," and checking is the exercise.


On reducing intake, if you decide to

NHS (nhs.uk) and NIAAA (niaaa.nih.gov) both publish free, practical, non-judgemental material on cutting down, including drink calculators and self-assessment tools. NIAAA's Rethinking Drinking site is genuinely good and free.

On alcohol use disorder specifically: NIAAA's treatment navigator, and national equivalents. ⚠️ Effective treatments exist — behavioural, pharmacological, and social. If any of §12.11's flags apply to you, these are the places to start, and starting there is not a bigger step than it sounds.


On sleep

Search for reviews on alcohol and sleep architecture. The pattern — reduced sleep latency, then suppressed REM and fragmented second-half sleep — is well characterized, and it's the effect most people notice first on reduction. Worth reading if the sleep argument is the one that would actually move you, which for many people it is.


A note on reading this topic

Alcohol has an unusual media environment: two loud positions, both overstating.

One tells you moderate drinking is fine or beneficial. Watch for: the J-curve cited without sick-quitter bias · Mendelian randomization unmentioned · resveratrol invoked · industry funding undisclosed.

One tells you any amount is dangerous. Watch for: Group 1 conflated with magnitude (Chapter 2 §2.6) · relative risks without baselines · the social and psychological benefits in §12.8 dismissed entirely · "no safe level" presented as "any amount is dangerous."

The tell that works on both: does the source give you an absolute number? A source that tells you the relative risk and not the baseline — in either direction — has not given you enough to decide with, and deciding is the whole point of this chapter.