Dose-related effect of smoking on mortality in critically ill patients: a multicentre cohort study

Purpose: It is uncertain whether smoking has an independent dose-related adverse effect on mortality in critically ill patients. This study assessed whether the intensity of smoking history, measured in pack-years, has a dose-related effect on mortality in critically ill patients. Methods: In this multicentre cohort study data were collected from six tertiary intensive care units (ICU) in Australia and New Zealand. Results: Of the 8,962 patients considered in the study, data on patients' smoking status and smoking history were available from 5,063 and 2,865 patients, respectively. Male gender, and chronic respiratory, liver and cardiovascular diseases were over-represented among smokers compared to non-smokers. Smokers had a higher risk of requiring mechanical ventilation and dying in hospital than non-smokers (10.7% vs. 6.7%, p = 0.001), particularly after emergency admission. Smokers also had a longer ICU stay than non-smokers (mean 3.2 days, interquartile range 0.8-3.2 vs. 2.8 days, interquartile range 0.8-2.9; p = 0.024). After adjusting for age, gender, elective surgical admission, severity of acute illness, and severe chronic illnesses, the intensity of smoking history remained significantly associated with the risk of dying in hospital. This was in a relatively linear fashion (odds ratio 1.08 per 10 pack-years increment, 95% confidence interval 1.02-1.15; p = 0.02). Further grouping of smokers into active smokers and ex-smokers, or including patients with unknown smoking status in the sensitivity analyses did not change the association between the intensity of smoking history and mortality. Conclusions: Smoking has a dose-related adverse effect on mortality of critically ill patients after adjusting for other confounders