Thread by @ProfDeepakKumar

I am a recent but huge fan of @Freakonomics podcast. I've also been enjoying @DrBapuPod by @AnupamBJena

Episode today was about #ArtificialIntelligence in #medicine and I want to share my concerns about the discussion on racial disparities in knee pain (starting at 19:00)

A 🧵

https://twitter.com/DrBapuPod/status/1494767699427270656

#1 - The suggestion that only pain that can be explained by pathology seen on a knee xray is "real, organic, genuine" and other pain is "in someone's head" is hugely problematic and detrimental to patients seeking care for #Chronicpain and providers.

nature.com/articles/s4159…

This is not aligned with the role of nervous system sensitization, psychological factors, and other mechanisms underlying the chronic pain experience.
Imp to note that #machinelearning model from xrays explained only 16% of variability in pain overall

@NatureMedicine @oziadias

#2- Interpretation that pain was "due to" or "caused by" pathology seen in knee xray is not appropriate. No causal inferences can be made here IMO. The AI model only showed that findings from an xray can explain some of pain variability.
Thoughts?? @PedoiaValentina @vkola_lab

#3- Most importantly.

Let's assume for a second that the knee pathology on the xray detected by the AI model is indeed causing the pain.

The study showed that something in the knee xray explained a large (43%) of the disparity in pain reported by Black and White patients.

It seems that this finding of the racial disparity in knee pain being explained by knee xray is being interpreted as some biological process (or "things") isolated to the knee and unrelated to "external factors" like "stress, mental illness, and just the burdens or everyday life"

This interpretation is what is most concerning. Before I share my thoughts, I want to admit I have matured and evolved in my understanding of how to think about data related to race in #clinicalresearch

Indeed, I would have accepted this interpretation as recently as last year

However, now (hopefully) knowing better, I want to explicitly state that race does not reflect any biological or genetic predisposition to disease. Race is a social construct designed to perpetuate #whitesupermacy

@ludriscoll

So, while Black patients DO have more arthritis on xray (visible or invisible to radiologists), this difference in pathology AND in the pain experience ARE due to complex and numerous external factors related to the experience of #systemicracism by minoritized racial groups

So "stress, mental illness, and just the burdens of everyday life" that are greater in Black patients with chronic pain due to individual and structural racism could lead to altered biology that may be seen as greater arthritis on xrays

These experiences and exposures over the life-course can increase the risk of arthritis (for example, due to greater obesity).

So, interpreting these data as "internal to the knee" and "external" is problematic.

#DoBetter

Also want to point out that just in ep #23 on placebo and nocebo effects (which I enjoyed btw), @AnupamBJena was open to the possibility of stress altering biology!

freakonomics.com/podcast/is-the…

I encourage @DrBapuPod to create an episode on race and health inequities from the lens of #CriticalRaceTheory as more than ANY other topic, this one is right "at the sweet spot of health and economics" !! 🙂

I also encourage @AnupamBJena to invite and elevate scholars from the minoritized communities who have done the work.

Any suggestions for Bapu? @rheum_cat #AcademicTwitter #HealthDisparities

I applaud the initiatives by @oziadias on developing #PrecisionMedicine tools that are anti-racist.

I want to end with an acknowledgement that I may be completely off the mark here. I assume #twitter will let me know!! 🙂

And of course, citations:

nejm.org/doi/full/10.10…

nejm.org/doi/full/10.10…

healthaffairs.org/doi/full/10.13…

jamanetwork.com/journals/jama/…

annfammed.org/content/annals…