|Weight or Smoking Classification||Mortality Rate per 10,000 person years†||95% confidence interval‡||Hazard ratio*||Range of hazard ratio|
|Normal Weight||17||16-18||1.0||reference group|
†The entire population was followed for 38 years during which time nearly 3000 people died. The death rates for each subgroup are expressed in person years. The person years are calculated by multiplying the total number of people by the number of years they were followed. So it was nearly 46,000 persons times 38 years = 1,748,000 person years. To work with convenient numbers, the death rate is expressed as the number of deaths in the group of interest per 10,000 person years.
‡The confidence interval defines the range of death rates that researchers are 95% certain includes the true death rate for that group (i.e., nonsmokers, overweight, obese, etc.)
*The hazard ratio provides a quantitative indication of the increased risk. The risk of death is doubled (2.14) in obese men as compared to normal weight men. Likewise it is doubled (2.11) in heavy smokers as compared to nonsmokers. The authors report that no statistically significant interaction between BMI status and smoking status was observed although the interaction between heavy smoking and obesity approached statistical significance. This failure to achieve statistical significance may be due to the relatively small number of individuals who died who were both obese and heavy smokers. In other words, the study was underpowered to detect the interaction.
Source of data: Neovius M, Sundstrom J, Rasmussen F. Combined effects of overweight and smoking in late adolescence on subsequent mortality: nationwide cohort study. BMJ 2009 Feb 24; 338: b496-504 doi: 10.1136/bmj.b496