Hello everyone,
I am analyzing a cohort data for the associations of macronutrients intake with the risk of disease using Cox Proportional Hazards regression analysis. My sample size includes 4500 subjects.
I examined each macronutrient individually using the residual method.
The model I used was:
HR= log(h0(t)) + β1Macronutrient + β2 total energy intake + Covariates.
However the results that I got had very narrow confidence intervals.
How I can handle such a 95% CI?
Sorry, Im not entirely sure what you’re asking? A narrow confidence interval just means that you can be very confident in the HR eatimate.
Thank you for your answer. That's right a narrow confidence interval means that I can be more confident in the HR estimate. But what I meant here if I got HR and confidence interval of 0.99 (0.99-1.00) it seems unrealistic estimate.
I suspect it could be to do with the units of your covariates e.g. 1 unit increase of your macronutrient may have such a small impact, thus the HR is close to 1 but still significant. If you divide your units by 10 for example then your HR would correspond to a change of 10 of the original units.
Thank you for your answer. It was really helpful. I applied it and got clear estimates.