Interpreting hazard ratios

#1
I'm currently studying interpreting hazard ratios. I thought I understood it but in one of the papers I'm studying something doesn't seem right.

The paper (ref below) is a cohort study into opium use and all-cause mortality, it's results are hazard ratios. They reported an adjusted hazard ratio of 1.86 (1.68-2.06) for those who used opium during the study period. Which makes sense. In subgroup analysis almost none of the other exposures (sex, ethnicity etc) had significant interactions with opium use.

However smoking tobacco did. The adjusted hazard ratio for smoking was 1.50 (1.27-1.77) for smokers, and 2.07 (1.83-2.35) for never smokers. In my summary of the paper I have included that "smoking cigarettes reduces the risk of all cause mortality amongst those who ingest/smoke opium". However this doesn't seem plausible to me, in biological terms. The authors of the study don't allude to this interaction or try to explain it so I'm a bit stumped.

I'm just wondering if I have correctly understood the hazard ratios? Or if something happens when you combine two risk factors which would actually lower the hazard ratio?


--------------------------------------------

Khademi, H., Malekzadeh, R., Pourshams, A., Jafari, E., Salahi, R., Semnani, S., Abaie, B., Islami, F., Nasseri-Moghaddam, S., Etemadi, A., Byrnes, G., Abnet, C. C., Dawsey, S. M., Day, N. E., Pharoah, P. D., Boffetta, P., Brennan, P., & Kamangar, F. (2012). Opium use and mortality in Golestan Cohort Study: prospective cohort study of 50,000 adults in Iran. BMJ (Clinical Research Ed.), 344. https://doi.org/10.1136/bmj.e2502
 

hlsmith

Less is more. Stay pure. Stay poor.
#2
I haven't looked at the paper, but you are talking about a multiplicative "statistical interaction", correct.