With over four-million-person years, the Nurses’ Health Study provides valuable insights.
“Is my rheumatoid due to something I’ve done, Doctor?”
Some of us have been in practice for many years, and my usual answer to this question was a resounding but comforting, “No”.
There was the data that suggested smoking interacted with genetic factors and some occupational associations (coal miners, for example) but these seemed rare in my patients and the usual advice was keep as active as you can, stop smoking if you are CCP positive (in particular) and maybe eat less saturated fat. Overall, I suggested that most of these (apart from smoking cessation, based on the CORRONA RA database) wouldn’t do very much.
However, in recent years there have been multiple papers on this issue which I had a vague knowledge of.
The authors of the featured paper have previously published individual papers on each of five risk factors, doing the original salami cutting exercise. Now, they decided to put them all together which would have been the best thing to do in the first place, but academics will be academics (even if the last paper would have had great impact if published first). This is a discussion I often have with my students – the first aim is to make a difference in a real sense, not just to your CV. To be fair, these studies are hard to do, as cohort studies of rheumatoid require millions of patient years of observation.
The study discussed here is from the Nurses’ Health Study, which is predominantly white females from the USA with assumed high levels of education. This may be an issue as RA does have a socioeconomic gradient, nevertheless the study meets Rothman’s criteria for generalisability.
The study is large, with over four-million-person years and over 1000 incident RA cases. Some things they can measure very accurately, such as BMI, smoking and antibody status. Diet and physical activity are assessed by questionnaire, which is pretty murky and makes these prone to regression dilution bias (i.e. the real association may actually be stronger than reported).
From previous papers, they identified five risk factors: BMI (healthy range), alcohol (one drink a day), activity (30 minutes a day), diet (best 40%!) and never smoking. These were all reduced to binary measures, which seem a bit arbitrary to me, especially for diet, and lessens the impact.
However, the results were surprising in that the having all five was associated with a much lower risk of RA (75% for seropositive and 58% for seronegative).
There was also a dose-response association, and the calculated population attributable risk (PAR) was 34%. This is the amount that is potentially reversible if everyone did the right thing. BMI (deleterious), alcohol (protective) and smoking were the main contributors, with little independent effect from diet and activity (maybe due to how they were measured).
This is the best data we are going to get on these issues, and I have changed my practice to include some discussion with my patients.
Hahn J, Malspeis S, Choi MY, Stevens E, Karlson EW, Lu B, Cui J, Yoshida K, Kubzansky L, Sparks JA and Costenbader KH (2022). Association of Healthy Lifestyle Behaviors and the Risk of Developing Rheumatoid Arthritis Among Women. Arthritis Care Res. doi.org/10.1002/acr.24862
Graeme Jones is professor of rheumatology and epidemiology with the Menzies Institute for Medical Research.
