The results of this population-based study show a decline in crude population AMI incidence and hospitalization in Girona Province after the 2006 Spanish partial smoke-free legislation enactment. This decrease was more pronounced among women, people aged 65–74 years and passive smokers. We also found a significant decrease in AMI mortality in women and people aged 65–74 years. However, no association was found between the partial smoke-free legislation enactment and AMI case-fatality.
The association between the implementation of a total smoke-free legislation and a reduction in hospitalization and population AMI incidence has already been described. Total smoke-free legislations have been estimated to decrease AMI hospitalization by 8–19% 
. A lower decrease in hospital admissions is expected after the implementation of a partial smoke-free legislation, as this type of law will still allow smoking in certain places 
. Our results show 11% decrease in AMI incidence and hospitalizations, similar to a recent report that showed 8.6% decrease in hospital admissions for AMI after a partial smoke-free legislation in Germany 
. One possible explanation our estimate is not lower than the one reported from meta-analyses from total smoke-free legislations could be the duration of the post-ban period. We have included a 3-year post-ban period, one of the largest to date, and it has already been shown, that the larger the post-ban period, the larger the reduction in AMI incidence 
Most of the previous studies that analyzed the effect of the implementation of smoke-free legislation, used hospital-based AMI registries. Only two Italian studies used population-based registries 
, with contradictory results: one reported a reduction in acute coronary events in Rome 
and the other showed no association in the Tuscany population 
. Gasparrini et al. 
pointed out that the inconsistency between studies could be partially related to differences in the definition of the statistical model. Estimation of the effect of the bans is sensitive to the definition of the statistical model used to analyze the data, specifically the inclusion of the AMI incidence rate trend and the assumption of linearity of this trend. To prevent this potential bias, we tested the linearity of AMI incidence during the analyzed period, included an adjustment for the linear trend as appropriate, and also adjusted for seasonality.
Previous studies have yielded contradictory results regarding the effect of smoke-free legislations on AMI incidence and hospitalization rates by sex. Some described a higher decrease in men 
, while others found a higher decrease in women 
or similar results in men and women 
. Our study supports a higher decrease in AMI incidence and hospitalization rates in women than in men. It should be noted that the population incidence of AMI in women of Girona is among the lowest rates in the world 
. The greater benefit in this population could be related to women's higher sensitivity to tobacco smoke compared to men. It has also been shown that smoking has a larger detrimental impact on myocardial infarction in women than in men 
. A plausible biological explanation for women's higher sensitivity to tobacco smoke is the alteration of lipid metabolism through the anti-estrogenic effect of smoking 
. Estrogens have a beneficial effect on LDL- and HDL-cholesterol metabolism providing a protective effect against myocardial infarction in women 
. However, women who smoke are relatively estrogen deficient and have decreased levels of HDL-cholesterol 
. Our age group results, showing a larger effect in people aged >64 years, do not concur with previous studies showing a stronger AMI incidence/hospitalization decrease in people aged <60 or <70 years 
. We think that several factors are contributing to this difference. First, the 2006 partial smoke-free legislation banned smoking in all indoor public places and workplaces but allowed exceptions in hospitality venues. People aged 35–64 years will probably spend much more time at the places where smoking was still allowed (restaurants, bars, discos, etc) than people aged 65–74 years. Second, smoking sums up to the other risk factors, which are more prevalent in people aged 65–74 years. People aged 65–74 years also contribute more to the total number of AMI events than the 35–64 years group. Thus, a reduction in SHS exposure in the older age group would have a more profound effect, as this age group is at higher risk, and a reduction in one of the more significant risk factors would avoid a large number of cases.
In accordance with previous studies 
, we observed a decrease in AMI incidence/hospitalization rates among passive smokers, while no effect was observed in the smokers subgroup. Spanish partial smoke-free legislation led to a reduction in SHS exposure 
, with a median decrease in nicotine concentration ranging from 60.0% in public places to 97.4% in private spaces, 96.7% in bars and restaurants that became smoke-free, and 88.9% in the no-smoking zones of venues with separate spaces for smokers 
. Moreover, significant reductions on SHS were seen at individual 
and population level 
. However, no changes were seen neither in salivary cotinine concentration nor in self-reported exposure to SHS in workers at hospitality venues where smoking was not totally banned 
. Reduction in SHS exposure has also been reported in other populations after a smoke-free legislation 
. All these data suggest that reduced SHS exposure was the fundamental underlying factor explaining the health effects of the smoke-free legislation.
Our results also showed an association between the Spanish partial smoke-free legislation and a reduction in AMI mortality, particularly in the population aged 65–74 years. Four studies have analyzed AMI mortality during a smoke-free legislation enactment, 3 in the USA 
and 1 in Spain 
, showing a reduction in AMI mortality only when a post-ban period longer than 2 years was considered 
. Our study analyzes this trend over the longest post-ban period reported to date and contributes to the evidence of an association between reduced AMI mortality and the enactment of a smoke-free legislation. We also found a non-significant decrease in AMI case-fatality after the partial smoke-free legislation enactment, indicating that the decrease in AMI mortality is mainly due to the decrease in incidence.
The effect sizes of the reported associations are very important from a public health perspective. The post-ban 18% AMI mortality decrease observed in the population aged 35–74 years is comparable with the reported effect of blood pressure control on coronary heart disease mortality in the 1988–2005 period in Spain 
To avoid the effect of including different definitions of AMI over time, we adhered to the new AHA/ESC definition algorithms 
and excluded 2001 events to avoid the effect of the gradual implementation of the new definition. In parallel we used the classical WHO-MONICA AMI definition 
, which ignores the troponin values. We observed that although the results are concordant, the effect of the 2006 Spanish partial smoke-free legislation was slightly higher when using the latter definition. This difference could explain some of the inconsistencies observed in previous reports.
Our study has several strengths. First, we used population-based data drawn from an AMI registry operating since 1990, which included all hospitalized cases and out-of-hospital deaths. Second, the availability of individual patient-level information such as smoking status permitted us to analyze subgroups of the population. Third, the present study is the first to present a parallel analysis using both AHA/ESC and WHO-MONICA AMI definitions. This approach increases the robustness of our results, bypassing the impact of using the AHA/ESC AMI redefinition alone, which could have introduced a bias. Finally, our data were collected with identical methods over a long pre- and post-ban period, increasing the stability of the estimations.
Our study also has a number of limitations that should be considered. First, due to the ecological nature of the study, no causal relationship can be inferred between the implementation of the 2006 Spanish partial smoke-free legislation and the reduction in AMI incidence and mortality. Second, unmeasured potential confounding factors, such as patient comorbidities and AMI case severity, other activities affecting smoking behavior, changes in AMI prevention and treatment strategies, and changes in air quality may have operated together with the partial smoke-free legislation introduced in 2006. Third, the study does not include direct observations on SHS exposure and consequently, despite the reported statistical associations, it is not possible to confirm a reduction in individual SHS exposure during the months following the ban enactment.