Present: Levi Bonnell, Nancy Gell, Juvena Hitt, Rocky Kelley, Charlie MacLean, Jen Oshita, Liliane Savard, Adam Sprouse-Blum, Connie van Eeghen (9)
1. Warm Up: Chilly day…
2. Levi’s manuscript on nonresidential destination and primary care patient outcomes: The audience will eventually be public health/preventive medicine/family medicine researcher type folks. The previous paper went in to Preventive Medicine.
a. I would like specific feedback on each section:
i. Intro: Clarity, strength of argument, writing, and setting up the intro in general. I left in a few comments on where I think I need to expand. Also, I just added the first paragraph so it’s a work in progress and needs citations still.
ii. Methods: Clarity. TMI/TLI (too little info). Do you want more info on the 4 different smoothing methods? Any other suggested methods? Thoughts on choosing the simpler piecewise linear regression over a fancier, less interpretable model?
iii. Results: What do you think? Robust enough without adjusting for covariates since we’re not trying to prove causality? Could the figures be improved?
b. Discussion
i. This is not a “geography/spatial analysis” population; this will affect the technical terminology
ii. Title alternative: association between nonresidential destinations and health outcomes…
1. Or “The relationship between x and y is nonlinear” and make the title a declarative statement
2. Is it important to identify PC patients? These patients have demonstrated access to PC, but not a factor in the analysis
a. The introduction focuses on the relationship, not the PC patient population
3. Keep “patients” rather than “adults” as the participants were drawn from a clinical environment
4. Keep MCCs as part of the description although consider dropping “multiple”; include “density of destinations” as well
iii. Introduction
1. This may be the first study on a broad population looking at destinations, BMI, and other measures of health. This message can be strengthened.
2. Last paragraph: we sought two aims: nonlinear relationship and BMI/other measures.
3. NRDs are not familiar, although the concept is accessible. Need examples of low/mid/high density environments are. What does that middle range look like?
a. Figures/aerial views of examples?
b. We are measuring walkable destinations, which approximates rural, suburban, urban categories. These were criticized in a previous manuscript review.
c. Use “15 destinations” as a way to establish categories
d. No familiarity with hectares, which is the typical measure. Keep this for methods; in the intro keep it more descriptive, e.g. “suburb with a Starbucks a mile away” or a “Tesla station” reference
e. Have a paragraph about why you chose this method, better than rural/urban and other indices for the following reason
iv. Methods
1. Identify number of records in original data set; report the number used in results
2. Should the Dun and Bradstreet explanation be included in this report?
a. Just explain that categories were developed by Levi et al in previous work
3. The multivariate analysis shows the effect is attenuated; still nonlinear but less of a relationship. How to share?
a. Is the goal to ask readers to consider NRD in policy decisions or to influence future research by scientists? Focus is researchers considering nonlinear relationships – which may change the journal to submit to.
b. Relationship is not just about walkability; there may be something important about patients with MCCs. Diverse populations interact differently across environmental characteristics.
c. This is useful to researchers: look for nonlinear relationships but it is also useful to the general population: think about how walkability affects patient health, and what drives that.
d. Walkability is the focus here, but why were they walking: food, work, school, recreation… We are making an assumption that all walking is equal. NRDs can be categorized further, but this is our level of focus now.
e. Chat comment: Maybe not this paper, but: Does density alone predict mental health? Does the accuracy of prediction improve if you include BMI and chronic conditions?
v. Relevance of Covid, as the data collection period.
1. Extraneous environmental factor; does not affect predictor. Unknown degree to which it affects the outcome.
2. Consider making a table for personal use to evaluate variables and the way Covid might have influenced them.
3. If these results are similar to the previous, pre Covid study, does this confirm lack of influence of Covid? There is a lot of regional variability, as well as variability among patient populations.
4. Present this is as a continuing story to support the relationships you have found.
5. Consider finding an author for an accompanying editorial; set this up ahead of time. May need a broader manuscript providing some validation of the NRD measure as a predictor.
a. Do this soon!
c. Results
i. These curves look very similar. What is going on in the 0 – 20 range of Establishments?
1. The point is that we don’t understand all the mechanisms taking place in “walkability.” Most studies have focused on 20+ categories
3. Next week: TBD
Posts
