Put on Those Flight Socks!

I am not part of the cult of randomized clinical trials: For many questions they are often inappropriate, poorly designed or both. But when the clinical trial method is skilfully applied to the right question, the results should be taken seriously. As I describe on Stanford’s SCOPE blog, that’s why I have started wearing compression socks on long-haul flights.

Author: Keith Humphreys

Keith Humphreys is the Esther Ting Memorial Professor of Psychiatry at Stanford University and an Honorary Professor of Psychiatry at Kings College Lonon. His research, teaching and writing have focused on addictive disorders, self-help organizations (e.g., breast cancer support groups, Alcoholics Anonymous), evaluation research methods, and public policy related to health care, mental illness, veterans, drugs, crime and correctional systems. Professor Humphreys' over 300 scholarly articles, monographs and books have been cited over ten thousand times by scientific colleagues. He is a regular contributor to Washington Post and has also written for the New York Times, Wall Street Journal, Washington Monthly, San Francisco Chronicle, The Guardian (UK), The Telegraph (UK), Times Higher Education (UK), Crossbow (UK) and other media outlets.

8 thoughts on “Put on Those Flight Socks!”

  1. Your reflections on the problems of randomized trial inclusion and exclusion criteria are very much to the point. I would value your opinion on a related issue having to do with methodology. Generally, it is assumed that all potential confounders, known and unknown, have washed out in the randomization. Adjustment for predictors of outcome would not be needed if this is truly the case. However, it also makes sense to do some covariate adjustment for known strong predictors of outcome. Sample size requirements may be reduced if this process improves the statistical efficiency of the analysis. It is a rare study which does this. Do you think it should be done more often?

  2. In seat yoga.

    I can think up several exercises:
    Pressing the knees together, raising the feet off the floor for a 30 count, clenching a “fist” with your toes, hugging one’s knees, etc.
    Someone should figure out the best ones to alleviate this particular malady.

    But beyond that, if the prevalence is as high as suggested: 1 in every 50 long flights, then human-scale design would demand a “stretching area” on long flight aircraft with passengers being educated, and put on a rotating timer, as a goad to get up and get some movement done. But of course that won’t get done as human scale loses out to the need to maximize profits (Perhaps as good a definition of American exceptionalism as one can conjure).

    Which of course, also explains why tomatoes don’t taste like tomatoes any more…
    And why airline food no longer exists as well…

  3. Ed: My understanding is that if there are known predictors of outcome, the randomization should be stratified on those variables, which produces greater statistical power with fewer observations. More generally, procedures like ANCOVA were developed for situations in which the groups differ despite randomization, even though they are now widely ab/used to try to create a randomization-like field of inference for observational data.

    1. Of course, you need lots of participants before you can stratify; otherwise, the numbers in each cell start to evaporate.

      But on the more general topic of restrictive entry criteria, it is interesting to note that the first RCT in 1948, comparing streptomycin plus bed rest vs. bed rest alone for pulmonary TB, started out excluding patients with cavitary disease. Later “this view was abandoned, as many otherwise suitable cases had large cavities.” They did not want old-standing disease with thick cavities, but then they looked at the films and decided that some cavitary disease was of recent development. In other words, they were flexible enough to broaden their inclusion criteria after the trial was underway.

      Not a bad example to follow.

  4. “However, it also makes sense to do some covariate adjustment for known strong predictors of outcome. Sample size requirements may be reduced if this process improves the statistical efficiency of the analysis. It is a rare study which does this. Do you think it should be done more often?”

    BTW – read a book on clinical trials, or Heckman or Pearl. There’s a lot of methodology developed for this [not always used], as well as quasi-experimental designs. I have about 400 pages of papers I mean to read this year.

  5. I wrap my legs with elastic bandages. Try explaining that to the brain dead TSA agents.

Comments are closed.