Most people lack the medical literacy to understand any of this. People don’t read their letters telling them to not eat before surgery, are they going to read the list of potential side effects of a medication? How many people read the information sheets that come with their prescription medication?
But there’s a potential risk of an allergic reaction to any drug you take, any food or drink you consume, even environmental substances - perfumes; hayfever is an allergic reaction to pollens. You don’t know you have one until you have the reaction. I didn’t know I was allergic to penicillins until I needed them for an infection and it turns out I am.
There’s really a risk vs benefit. If you have a brain tumour you need contrast to assess the type of tumour, its growth, if it’s a glioma whether it’s transformed and so on. If someone is being given contrast it is going to change their clinical management.
It seems an odd fixation of just MR contrast when the same could be said of all drugs. Does your doctor/surgeon go into the minutiae of all drugs and possible consequences? By this line of thinking, saline is not without risks, should they go into depth about that?
People already poorly retain information or even comprehend it at appointments or interventions, is there any point adding more burden onto their attendances?
The article linked was very scant on references too. A majority of his references per article are one or two papers and health websites, such as Healthline.
On the satiety page [1] him and another scrape MyFitnesPal diary entries and uncritically label diaries where the intake exceeded desired calorie amounts as having low satiety, and vis versa. It somewhat ignores that people may have not logged all of their food for the day; I would have expected to see controls for that, and maybe ensuring the data was coming from regular loggers instead of incidental ones. The actual method of analysis is not well-described.
Again, when claiming to be driven by data:
- In the Holt paper, there are 41 people are broken down into 6 groups, so ~11 people per group
- When you look at the the analysis, the SEM can correspond ~25-30% of the mean value; when you consider this to be ~11 people, when the I score for popcorn was 154 +/- 40 the veracity/reliability of the data for founding all of his arguments starts to become challenging
- The Holt data uses a scale looking at six different qualitative factors of the food/eating experience and constructs a score from this and satiety perception after 2hrs and uses bread as a baseline
- A lot of his graphs are missing proper labels
- Many have lines where the data's source is poorly indicated
- He uses nebulous terms such as satiety and hunger without it being effectively operationalised (or having a meaningful scale)
- The "The Optimising Nutrition Satiety Index" graph omits most of the labels and does not provide the data
- The same graph uses a linear correlation, but has a curved line and does not provide a p-value
- I managed to perform a partial reconstruction through. I found their SI data [2] and hovered over a lot of data points (found 26 of them which took a fair bit of time - you can't export it) and combined that with the Holt data. My graph approximates theirs relatively well [3]
- From my tentative analysis from a one tailed Pearson, I found: n=26, r=.631, r2=0.399, p<0.001, 95% CI (0.370, 1)
- I'm a bit sceptical that their r2 is in fact 0.6031 when the one I found was 0.339. I wonder if they are using the r value which I found to be 0.631.
- To be sympathetic and assuming the r was misstated (and my r would change with additional data), they have managed to demonstrate a moderate significant correlation with reasonable effect size, between Holt's measure and their own.
- However, Holt's uses a significantly different methodology to his, and to that end I fundamentally question if they are measuring the same thing as to justify his claims, such that the correlation could be co-incidental.
- I don't think it would be unreasonable to expect meaned data from ~11 people (and 41 data points) to have some form of relationship with data from 9000 (or 587187 days of entries).
- A majority of the article relies on the reader to look at a graph and draw an inference (without labels or meaningful information
- Interestingly the Satiety-Hunger axis is reversed on their simple diagrams compared to what the analysis uses
I think that when they consider themselves to be data analysts, the analysis has to be of a high-standard. Sloppy presentation, communication, withholding of data/methodology, reliance on one or two articles to substantiate claims (often times very old articles) generally leaves a lot of room to question their findings/claims.