Cox Regression

Hazard ratios and the proportional hazards idea

Back

Model intuition

h(t | X) = h₀(t) * exp(βX) HR = exp(β)

The hazard ratio (HR) compares instantaneous risk of event between groups. HR=1.5 means 50% higher hazard at any time t (if proportional hazards holds).

Interactive: what does HR do to survival curves?

We draw illustrative survival curves from a simple exponential baseline. This is a teaching visualization, not a Cox fit.

0.31.503.0
0.010.060.20

Survival curves (illustration)

If HR>1, survival drops faster for the exposed group.

Real Dental Scenario

Implant Survival: Smokers vs Non-Smokers

A study followed 240 patients with dental implants over 10 years. Researchers found that smoking was a significant predictor of implant failure using Cox regression.

Key Finding: HR = 2.1 for smoking

This means smokers have 2.1 times the hazard of implant failure at any given time compared to non-smokers, after adjusting for age, bone density, and implant type.

Dental example

Cox regression can model time-to-implant failure with predictors like smoking, diabetes, bone density, and clinician experience, reporting adjusted hazard ratios.