So can someone follow along with me to see what I'm missing:

NPR: 8 drinks a week for women

Why?

NPR: "To find out how the CDC came up with them, we talked with Lela McKnight-Eily, a health scientist at the CDC who is the lead author on the new study."

New study:

CDC analyzed Behavioral Risk Factor Surveillance System (BRFSS) data from a question added to surveys in 44 states and the District of Columbia (DC) from August 1 to December 31, 2011, about patient-reported communication with a health professional about alcohol. Elements of ASBI are traditionally delivered via conversation. Weighted state-level prevalence estimates of this communication were generated for 166,753 U.S. adults aged ≥18 years by selected demographic characteristics and drinking behaviors.

Results: The prevalence of ever discussing alcohol use with a health professional was 15.7% among U.S. adults overall, 17.4% among current drinkers, and 25.4% among binge drinkers. It was most prevalent among those aged 18–24 years (27.9%). However, only 13.4% of binge drinkers reported discussing alcohol use with a health professional in the past year, and only 34.9% of those who reported binge drinking ≥10 times in the past month had ever discussed alcohol with a health professional. State-level estimates of communication about alcohol ranged from 8.7% in Kansas to 25.5% in DC.

So… NPR says "two drinks a day are too many" because the CDC added 1 (one) question to an intake form for 5 (five) months in 2011 and came to the conclusion 1 in 4 "binge drinkers" talk about alcohol with their doctors, 1 in 6 "regular drinkers" and about 1 in 7 "non drinkers."

First thing: Okay….

Next thing: This isn't what we're supposedly talking about.

The findings in this report are subject to at least five limitations. First, BRFSS data are based on self-report and dependent on respondent recall of dialogue with a health professional, which can vary based on the time since the patient's last visit or other factors that could have affected patient recall, thus resulting in underreporting. Second, respondents were asked to report only whether they "talked with" a health professional about their alcohol consumption, not whether they reported their alcohol consumption in some other manner (e.g., on a patient history form) or if they were actually screened or received an intervention.

So, li'l story. Back in 2000 or so I went to see a doctor for a check up. First time I'd been in years, figured it was a good idea. Strange asian man, part of the UW health network, done in a strip mall next to Costco. And it was the damndest thing:

"Raise arm… don't do drug."

"Turn head, cough: always wear seat belt."

"Say aaah: always wear condom for sex. Don't get AIDS."

A couple years later I was going through records and mentioned this bizarre visit with my girlfriend, who had been a high-up software architect in the insurance industry. She laughed at me.

"Well, of course he did. Every single one of those PSAs is a box his secretary can check off on the form they're sending your insurance company. He's probably making thirty bucks for telling you to use a condom."

So you'll excuse me if I'm skeptical of this study's intended results, let alone NPR's projected results. For one thing, if their question isn't even granular enough for "did your doctor bilk your insurance company out of $30 by telling you not to drink too much" I'm unconvinced it's granular enough to be useful. One question on one form for five months? Puhhhleaze.

But to then extrapolate it out to "hey, we found a study that contraindicates every other study we've been trumpeting for the past ten years"?

Yeah, not buyin' it.