Latest Twitter Threads by @EAllen0417 on Thread Reader App

Jan 31, 2018 • 7 tweets • 4 min read

Chanced upon an interesting study from Sweden today. Sweden had some of the highest CVD mortality in the 1970s, so programs were put in place to decrease cholesterol. They were successful at getting reductions in serum cholesterol by decreasing SFA primarily in the 80s and 90s.

In 2004, the #LCHF movement got a lot of press and Swedes seemed to be convinced by social media campaigns to abandon their dietary guidelines and decrease carbohydrates and increase fats. You can see that they did exactly that from the graphs. What happened?

Dec 21, 2017 • 7 tweets • 2 min read

Some final thoughts on S-T & K's meta-analysis on SFA and CVD. When this was published back in 2010, headlines and news articles repeatedly called it "new research." The same canards were recycled when Chowdhury published a similar study 4 years later.

Of course, none of that is true. These were not new studies. Medicine is a human endeavor and people who are told there is "new research" that exonerates SFA should be getting actual new research.

Dec 20, 2017 • 8 tweets • 3 min read

Our final, non-duplicate study in S-T and K's meta-analysis is another study of strokes out of Japan. The Adult Health Study (apparently rated R), looked at a single category of cerebral infarction in Japanese from Hiroshima and Nagasaki.

The answer to the question you have in your mind right now is, "Yes, they were all exposed to the atomic bombs." So this is not your typical run-of-the-mill nutritional epidemiological study. What it's doing in this analysis is really hard to justify IMO. stroke.ahajournals.org/content/35/7/1…

Dec 19, 2017 • 11 tweets • 3 min read

I want to take a final census of the studies we've looked at before doing the stroke studies from the S-T & K meta-analysis. Adopting the most favorable reading of all studies (studies with no overall significant results) it's 10-6 harmful vs. no harm.

None of the studies of CHD and SFA showed benefit to SFA, and when we read it critically, including all studies that showed harm from a subgroup, we get 14-2 harmful vs. no harm. I want to point out that all the studies excluded people with existing CHD.

Dec 18, 2017 • 7 tweets • 3 min read

Our final paper from ST-K's series on heart disease is the Strong Heart Study published in 2006 by Xu. This was a study of 3k people in Arizona, Oklahoma and the Dakotas. Conducted through the IHS, it used 24 hour dietary recall.

The patients were followed for about a decade, those with angina or known MI in the past were excluded. In patients under 60, there was a strong association between SFA and heart disease, that disappeared in those over 60. Trans fats had no association in all groups.

Dec 15, 2017 • 6 tweets • 3 min read

Next up we have the Baltimore Longitudinal Study of Aging. This looked exclusively at 501 men free of disease at the outset that were followed for 18 years. Seven day diet records were recorded at multiple (up to 7x) intervals in the study.

The BLSA took so long that the nutrition calculations method were changed over the course of the study, but the investigators were scrupulous and unified the methods before the final analysis. The most interesting choice they made was dividing dichotomous variables into binaries.

Dec 14, 2017 • 5 tweets • 2 min read

Today we look at the Nurses Health Study. This is an observational trial with ~80k female nurses. FFQs were done biennially and patients with CVD, DM and hyperlipidemia patients were excluded. They were followed for eighteen years.

For women in this study, total fat dropped from about 40% of calories to about 30% of calories over the time of the study. PUFA was protective of CHD, MUFA, TFA and SFA all were associated with increased CHD.

Dec 13, 2017 • 7 tweets • 3 min read

Today we take a look at the Malmo Diet and Cancer Study published in 2007. This study was primarily to study cancer risk factors and looked at 28k people in Malmo, Sweden. They were followed for 5 years and food info was from 7 day diet entries.

The study did not record trans fat as a separate category, so most of the trans fats were counted as MUFA or PUFA. This is a pretty profound confounder and makes the study almost uninterpretable. journals.sagepub.com/doi/abs/10.109…

Dec 12, 2017 • 13 tweets • 4 min read

Before we start our look at the next article in S-T & K's meta-analysis, I want to briefly point out a concept. When a study is looking at an association in epidemiology, there are three possible outcomes.

First, the study can show a positive association, the more you do of something, the more likely you are to get a specific condition. This is true of being a chimney sweep and scrotal cancer, for instance. oem.bmj.com/content/40/4/3…

Dec 11, 2017 • 7 tweets • 2 min read

Today we look at the Health and Lifestyle Survey by Boniface and Tefft published in 2002. This study looked at ~2600 men and women. The methodology was a single interview with follow up. Patients with existing CVD were excluded.

The study showed a strong and significant effect on CHD mortality in women, but none in men. The pooled analysis was not mentioned in their results. The average SFA consumption was 414 calories per day in the non-excluded patients. This is 20% of a typical day's calories.

Dec 8, 2017 • 8 tweets • 3 min read

Today's study is the alpha-tocopherol beta-carotene cancer prevention study done in Finnish men. The primary purpose of the study was not to determine dietary information regarding CHD but to test whether supplemental vitamins prevented cancer in smokers.

The study looked at ~22k men, all of whom smoked at the start of the study. The study found no association between SFA and CHD, but did find positive associations for omega-3s and for TFA which were signficant. Fiber and omega-6s were protective. ncbi.nlm.nih.gov/pubmed/9149659

Dec 7, 2017 • 5 tweets • 2 min read

The following study in our consecutive look at S-T & K's SFA exoneration is Mann's prospective cohort of health-conscious individuals in the UK. This study looked at ~10k adults and followed them for 13 years after a single FFQ.

The study compared vegetarians and non-vegetarians to find out what the CHD rates were. The vegetarians were lacto-ovo vegetarians or semi-vegetarians, there was no subgroup that was vegan. ncbi.nlm.nih.gov/pmc/articles/P…

Dec 6, 2017 • 5 tweets • 2 min read

Next up on our voyage is the Lipid Research Clinics Prevalence Follow Up Study (LRCPFUS). Nothing too much to see here and I was unable to get a .pdf for it. If someone sends one I can certainly get more info. from it.

This is a Canadian study that took FFQs from people at baseline and followed ~4500 people for twelve years. The results were interesting if not overwhelming. sciencedirect.com/science/articl…

Dec 5, 2017 • 9 tweets • 3 min read

The next study on our tour of S-T and K's paper that "exonerated" SFA in coronary artery disease is an analysis of the data collected in the health professional's follow up study done at Harvard.

Once again, we see a homogenous population with very small differences between groups. In this case, US physicians, followed for six years. Dietary records were biennial FFQs. I remember one of my professors was in the study when I was in medical school. scholar.google.com/scholar_url?ur…

Nov 29, 2017 • 9 tweets • 3 min read

The next in line is a genuinely anomalous study, the Caerphilly study. It's an observational study of ~2500 men in Wales that were followed for five years starting in 1979. This study showed a remarkable trend. Those who ate the least had the most CHD.

The finding was that total energy intake was the only statistically significant predictor of heart disease, and it had an inverse relationship with CHD. Yes, those who ate the fewest calories had the highest risk. This would overturn most of what we know about nutrition if true.

Nov 28, 2017 • 6 tweets • 2 min read

Following in our sequence from the S-T & K paper on SFA, the next paper is the Framingham study nutritional analysis from Posner. Of note, Framingham did not initially include a dietary component.

The study began in 1948 and the first food interviews didn't happen until 1954. These were a combination of 24 hour recall and FFQ. This study, published in 1991 controlled for serum cholesterol, yet still found SFA was linked to CHD. jamanetwork.com/journals/jamai…

Nov 27, 2017 • 5 tweets • 2 min read

The next study in my sequence of studies used by S-T & K is the Boston Diet Heart Study. Published in 1985, the study looked at observations on men living in Ireland, men born in Ireland living in Boston and Irish men born and living in Boston. Echoes of Ni-Hon-San, right? Except that unlike Ni-Hon-San, there were no differences between the groups at all. The study was based on a single dietary recall that was then scored using the Hegsted and Keys scores. The study was unremarkable, showing exactly what you would expect.

Nov 22, 2017 • 11 tweets • 4 min read

This will be the second study in the list from Krauss and Siri-Tarino's dairy-funded study on saturated fat (first below). This one is the Honolulu Heart trial. S-T & K used the 1984 paper but much more has been published on this dataset.

Honolulu Heart, part of a larger study called the NiHonSan study looking at Japanese people living in Japan, Hawaii and California determined what risk factors were present that were associated with heart disease. For reference, Framingham showed 12/1000 risk of CHD.

Nov 21, 2017 • 9 tweets • 3 min read

Since there is a modicum of interest, when time allows I'll be going through the studies referenced in Krauss & Siri-Tarino's paper. These studies don't have uniform results and it's important to allow for heterogeneity.

However, as it relates to heart disease specifically, it is important to understand the normal distribution. If something is truly mildly harmful, we would expect results of studies investigating it to show a few beneficial results, mostly harmful effects and a few neutral.

Share this page!

Enter URL or ID to Unroll