Quantification of the Effect of Smoking and Smoking Cessation on Lipid Parameters – A Metaanalysis

Abstract

Conference ACE Meeting 2013 Louiville, KY, USA 21st-24th September The Medline database (Pubmed) was searched for studies that evaluated the relationship between smoking or smoking cessation and lipid parameters which included: HDL-C (including its sub fractions HDL-C2 and HDL-C3), apolipoproteins (Apo A-I, Apo A-II, Apo B), Lip (a) and TG. The search was performed between September 28th and April 10th 2013 using the following key words: “smoking”, “smoking cessation”, “quitting”, “apolipoprotein”, “Lip (a)”, “High Density Lipoprotein”, “triglyceride”. Selection of articles was further limited to those written in English and considering human populations. To identify other available studies, the reference lists of the publications obtained through the original search were checked for any additional articles. Inclusion Criteria • Case control or cohort studies (observational and experimental studies) • Adult human populations were studied • Measurements of HDL-C, TG or Apolipoproteins by exposure with the following measures available: mean lipid levels by group, SD or SE (of the mean), sample size per group or with enough information to allow for the calculation of mean and SD. • Studies published after 1970 (inclusive) Exclusion Criteria • Review articles, case reports, articles or editorials • Reports with incomplete data which could not be incorporated into Revman 5.1. • Duplicate publication of the same study • Studies performed in diseased populations • Unknown time of quitting for cessation studies Statistical Analysis To quantify the effects of smoking and smoking cessation on lipid levels, pooled mean differences between smokers and non-smokers (when assessing effects of smoking on lipids) or between quitters and their baseline measurements (when assessing the effects of quitting smoking on lipid parameters) and 95% confidence intervals (95%CI) were calculated using the fixed-effects model in Review Manager version 5.0 (Cochrane Collaboration, Oxford, UK). The degree of heterogeneity between the study results was tested by the inconsistency statistic (I2) (Higgins et al 2003), when the between study heterogeneity was high (I2>75%) the random effects model was used to correct for heterogeneity and the results compared to the fixed effects model. Funnel plots were used to evaluate publication bias (Macaskill et al 2001). Search Results A total of 1293 studies were identified through the Pubmed search and the reference list check yielded 81 studies. A detailed description of the data extraction is shown on Figure 1.