Physicians’ detection of smokers: A multilevel analysis of the Paris Prevention in General Practice survey

Objective

Although many studies have examined the mention of smoking status in case files, a more important issue in clinical practice is physicians’ ability to identify smokers. We sought to analyze physicians’ detection of smokers according to characteristics of patients (social especially) and physicians.

Methods

In 2005–2006, 59 randomly recruited general practitioners from the Paris metropolitan area enrolled every man aged 35–64 years seen during a two-week period. Physicians’ detection of smokers was analyzed in a logistic mixed model that considered patient (occupational class, education, income, and social integration) and physician (general demographics, practice organization, smoking control practices, personal smoking status) characteristics.

Results

Of the 1096 participating men, 35% smoked. The detection rate (55%) did not vary between physicians. Detection was better for men with low social integration (versus high or intermediate-high, OR = 8.33, 95% CI = 2.23–31.1) and low income (< 1000 versus ≥ 3500 €, OR = 2.88, 95% CI = 1.00–8.25) and for physicians in practice for less than 20 years (OR = 0.43, 95% CI = 0.23–0.82) and ex-smokers (versus never-smoker, OR = 2.97, 95% CI = 1.45–6.01), independently of patient age, physical activity, chronic disease and length of the patient-physician relationship.

Conclusions

The better detection observed for smokers at the bottom of the social scale and among newer physicians are positive factors that merit confirmation.