Impact of smoking on procedural outcomes and all-cause mortality following acute myocardial infarction: A misleading early-stage pseudoparadox with ultimately reduced survival

Background

Smoking has conflicting results on outcomes following acute myocardial infarction (AMI). We evaluated the independent influence of smoking status on patient outcomes.

Methods

We included patients with AMI undergoing invasive coronary angiography with available self-reported smoking status. The incidence of death of any cause was evaluated during a median follow-up of 1.14 years (range 0.36–3.40 years). Association between smoking status and long-term mortality was evaluated using multivariable adjusted Cox regression analysis.

Results

From 1612 AMI patients (aged 65.7 ± 13.3 years, 72.1 % male), 378 patients (23.4 %) were current-smokers, 311 (19.3 %) ex-smokers, and 923 (57.3 %) non-smokers. Compared with non-smokers, current-smokers were younger (68.5 ± 13.0 vs. 58.6 ± 12.5, p < 0.0001) and more frequently presented with STEMI (21.6 % vs. 35,4 %, p < 0.0001), while ex-smokers with similar frequency of STEMI-manifestation as non-smokers (22.5 %, p = 0.79) constituted an intermediate-group in terms of age (65.8 ± 11,6 years). Although smoking status was not significantly associated with long-term survival in unadjusted-analysis, active-smokers had 56 % higher long-term mortality than non-smokers when adjusting for age, gender, medications and other traditional risk factors, whereas ex-smokers possessed comparable survival probability (current-smokers: 1.56[1.14–2.14], p = 0.006, ex-smokers 1.16[0.84–1.59], p = 0.37). Current-smokers had unadjusted lower NT-proBNP and modestly higher absolute in-hospital left ventricular global longitudinal strain (LV GLS) values that did not differ among groups after the same adjustments (NT-proBNP: −0.08[-0.31; 0.15], p = 0.5, LV GLS: 0.65[-0.26; 1.55], p = 0.16).

Conclusion

Active smoking is associated with increased adjusted long-term mortality, earlier onset and more frequent manifestation as STEMI, compared to non-smoking. Comparable adjusted results for LV GLS and NT-proBNP between groups support the presence of the pseudoparadox.