Article Text

Download PDFPDF

Decrease in mortality rate and hospital admissions for acute myocardial infarction after the enactment of the smoking ban law in São Paulo city, Brazil
  1. Tania M O Abe1,
  2. Jaqueline Scholz1,
  3. Eduardo de Masi2,
  4. Moacyr R C Nobre1,
  5. Roberto Kalil Filho1
  1. 1Medicine Faculty, Heart Institute, University of São Paulo, São Paulo, São Paulo, Brazil
  2. 2Municipal Health Secretary, Municipality of São Paulo, São Paulo, São Paulo, Brazil
  1. Correspondence to Tania M O Abe, Heart Institute, Medicine Faculty, University of São Paulo, Rua Dr Eneas de Carvalho Aguiar 44—1 andar Bloco 2–São Paulo, SP CEP 05403-000, Brazil; drataniaogawa{at}gmail.com

Abstract

Background Smoking restriction laws have spread worldwide during the last decade. Previous studies have shown a decline in the community rates of myocardial infarction after enactment of these laws. However, data are scarce about the Latin American population. In the first phase of this study, we reported the successful implementation of the law in São Paulo city, with a decrease in carbon monoxide rates in hospitality venues.

Objective To evaluate whether the 2009 implementation of a comprehensive smoking ban law in São Paulo city was associated with a reduction in rates of mortality and hospital admissions for myocardial infarction.

Methods We performed a time-series study of monthly rates of mortality and hospital admissions for acute myocardial infarction from January 2005 to December 2010. The data were derived from DATASUS, the primary public health information system available in Brazil and from Mortality Information System (SIM). Adjustments and analyses were performed using the Autoregressive Integrated Moving Average with exogenous variables (ARIMAX) method modelled by environmental variables and atmospheric pollutants to evaluate the effect of smoking ban law in mortality and hospital admission rate. We also used Interrupted Time Series Analysis (ITSA) to make a comparison between the period pre and post smoking ban law.

Results We observed a reduction in mortality rate (−11.9% in the first 17 months after the law) and in hospital admission rate (−5.4% in the first 3 months after the law) for myocardial infarction after the implementation of the smoking ban law.

Conclusions Hospital admissions and mortality rate for myocardial infarction were reduced in the first months after the comprehensive smoking ban law was implemented.

  • Secondhand smoke
  • Surveillance and monitoring
  • Public policy
  • Environment

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

It is well known that passive smoking increases the risk of heart attacks. In the last 2 decades, many cities and some countries, motivated by the WHO Framework Convention on Tobacco Control, approved partial or total smoking ban laws, contributing to the decline in passive smoking. The main benefit of smoking ban laws is the reduction of passive smoking, which is associated with an increase in the risk of myocardial infarction of 30–60%.1–5 Although this risk might seem disproportionately high, it is consistent with laboratory evidence of increasing platelet aggregation and changing endothelial function (observed within 30 min of exposure to tobacco smoke), which can increase the risk of coronary heart disease and trigger acute coronary events.4–8 Concerning current smoking, previous studies showed benefits of smoking ban laws and other tobacco control policies in reducing current smoking.9–12

In most of the world, a decline has been observed in rates of hospitalisation for myocardial infarction. The decline prior to the smoking ban laws was attributed to improvements in the treatment of risk factors for myocardial infarction and implementation of other preventive measures. After the laws, the pre-existing trend of a decline in rates of hospitalisation for myocardial infarction became more evident, especially in small cities and localities;13–15 in larger cities and countries,16 ,17 the reduction of the rates was less significant—other factors such as migration make population control more difficult, which may have influenced these rates.

In this scenario, Uruguay, which has ∼3.5 million inhabitants, was the first Latin American country to implement a comprehensive smoking ban law, in 2006, and to corroborate these data. They observed a decrease of 22% in the admissions for acute myocardial infarction in the first 2 years after the enactment of the law.18

In Brazil, in 1965, the obligation of having warnings about the dangers of smoking on the cigarette packs was introduced. In 1996, some restrictions to cigarette advertising were created and a partial restriction law to not smoking in hospitals, public repartition and some public places were established. Since 2000, the restrictions became more severe.

São Paulo is a city of more than 12 million inhabitants. It is the most populated city in the southern hemisphere and the 14th most global city in the world. It was the first city in Brazil to enact a comprehensive smoking ban law.

The Sao Paulo smoking ban law prohibited the use of cigarettes and other tobacco products in closed and semiclosed places, public and private, with the exception of residences, places of religious worship where smoking is part of the ceremony and sites designated for the consumption of tobacco products.

To evaluate the consequences of this law, we conducted this study in two parts. The first part was published in 2010,19 when we analysed the close effect of the implementation of the smoking ban law in São Paulo. We reported a significant reduction in the indoor levels of carbon monoxide (CO) after the law in collective environments, such as restaurants, bars and nightclubs, without any significant change in the level of CO in the atmospheric air (outdoor), proving that the law was working. The same occurred with the level of CO in the exhaled air of workers in these same places. These measures were made before and 3 months after the enactment of the law.19

In this second phase of the study, we analysed the rates of hospitalisation and mortality for acute myocardial infarction before and after enactment of the law, in August 2009, in São Paulo city. We also analysed data for air pollutants, temperature and humidity of the air and other factors that could contribute to changes in these rates.

Methods

The Ethics Committee for Research Project Evaluation (CAPPesq) of the Hospital das Clinicas—School of Medicine, University of São Paulo approved this study. It is a time-series study, and we acquired data from January 2005 to December 2010. The law is dated from August 2009, and we acquired data 4 years before the law went into effect to estimate the behaviour of myocardial infarction time series, for hospital admission rate and for mortality rate.

Study sample

Data concerning hospital admissions for acute myocardial infarction were collected from DATASUS, the administrative database of the Sistema Único de Saúde (SUS; Brazilian Public Health System). It includes data from all SUS hospitals of the city, but not includes data from private hospitals. SUS system is responsible for about 55% of the medical coverage in Sao Paulo city, according to 2010 Report of Sao Paulo Municipal Health Department.

Mortality data were collected from Mortality Information System (SIM). It includes data from all the city, including public and private hospitals and deaths occurred out of the hospital.

We used the hospital admission diagnosis or death diagnosis, according to International Classification of Disease, 10th Revision (ICD-10), coded I21 (acute myocardial infarction), I22 (subsequent ST elevation (STEMI) and non-ST elevation (NSTEMI) myocardial infarction, I23 (current complications following STEMI and NSTEMI myocardial infarction) and I24 (other acute ischaemic heart diseases). The data were collected in a monthly basis from January 2005 to December 2010, and included residents of São Paulo city.

The total hospital admission for all diseases in SUS hospitals was also collected and called ‘total hospital admission’.

Data concerning the total population of São Paulo city were provided by the Data Analyses State System (SEADE) Foundation, which is based on the Geography and Statistics Brazilian Institute (IBGE) decennial census. We estimated the population between the two decennial censuses using geometric progression analyses.

Pollutant data

Data related to environmental variables and atmospheric pollutants—maximum and minimum temperature, air relative humidity, maximum CO concentration, nitrogen dioxide (NO2), sulfur dioxide (SO2), ozone (O3) and particulate matter (MP10)—were provided by CETESB (Environmental Company of São Paulo State) database. The data were acquired from January 2005 to December 2010, on a monthly basis.

These data were used to estimate models and to control confounding. Because of multicolinearity among the variables and the biologic plausibility of them related to acute coronary disease, the variables used in the models were CO, minimum temperature and air relative humidity.

Statistical analysis

The Autoregressive Integrated Moving Average with exogenous variables (ARIMAX) method was used to analyse the effect of the smoking ban law, modelled as a dummy variable, in the mortality rate and hospital admission rate data for myocardial infarction. The ARIMAX models were also adjusted to other parameters, including ‘total hospital admission’, CO, minimum temperature and air relative humidity. The ARIMAX method allows to estimate lag effects of input series and to forecast output series, as a function of a linear filter of the input series (transfer function) and of the noise (ARIMA filter) and by controlling for the autocorrelations. It enables us to compare the predicted rate of hospital admission and mortality with the real observed rate.

Transfer function models were estimated using the ARIMA command for ‘Data Analysis and Statistical Software’ Stata/LP, StataCorp, V.10.1.

We also applied Interrupted Time Series Analysis (ITSA) integrated into ARIMAX to compare the series trend before and after the law and to verify changes in the level in the moment of the law. The time-series intervention models were conducted using the ITSA command in Stata/MP, StataCorp, V.14.20

Construction of the ARIMAX models

The modelling process followed the Box-Jenkins methodology. First, identification of the best model was performed based on autocorrelations and the presence of unit roots. The autoregressive orders p, moving average q and their corresponding seasonal orders P and Q were identified by the observation of the autocorrelation and partial autocorrelation functions. The presence of unit roots was tested using the augmented Dickey-Fuller test and Phillips-Perron test. Prewhitening and the cross-correlation function were performed to identify lag correlation order between input and output series.

Thereafter, transfer function models were constructed. In ARIMAX modelling process, environmental series (minimum temperature, air relative humidity) were inserted first. Thus, atmospheric pollutant series (CO) were input, followed by the intervention series. During the entire process, the model was adjusted to identify the series and their lags that could better explain the rates of mortality and hospital admission for myocardial infarction. Because of the multicollinearity among the pollutant series, we chose the series that provided the best prediction of the event under study and which have biological plausibility.

Two models were estimated in this process. First, a complete model was estimated for the entire period of the study, to evaluate the effect of the intervention (smoking ban) in the behavior of the series. Then, a second model was estimated using only the pre intervention data, and creating a forecast model for “future events”. Then, it was possible to compare the real event rate with the forecasted event rate.

Construction of ITSA-ARIMA models

In the ITSA analysis, the Prais-Winsten and Cochrane-Orcutt method was used. This method uses the generalised least-squares method to estimate the parameters in a linear regression model in which errors are assumed to follow a first-order autoregressive process.

At the end, a more complex ARIMA model was incorporated to ITSA procedure.20 The model's accuracy was verified by residue correlogram analysis, which has autocorrelation and partial autocorrelation functions, and by the Q test of Ljung-Box. The residue normality was tested by the Shapiro-Wilk test. The best model was chosen based on the Akaike Criterion Information, the mean prediction error percentage and the residual SD. In the entire analysis, 5% significance level was adopted.

Results

The number of monthly hospital admissions and deaths for myocardial infarction is shown in tables 1 and 2.

Table 1

Monthly number of hospital admissions for myocardial infarction, SUS, city of São Paulo, Brazil, January 2005 to December 2010

Table 2

Monthly number of deaths for myocardial infarction, city of São Paulo, Brazil, January 2005 to December 2010

The model estimated using ARIMAX for mortality rate due to myocardial infarction was adjusted for total hospital admission, minimum air temperature and maximum CO level (table 3). This model showed an immediate and a long-term effect of the law in the mortality rate (ω0t−0=−0.52, p<0.001), reducing it in almost 12% per month. This effect persisted for all the pos-intervention period (17 months) (figure 1). According to this model, 571 myocardial infarction deaths were avoided after the law.

Table 3

Transfer function model for mortality rate due to myocardial infarction, city of Sao Paulo, Brazil, January 2005 to December 2010

Figure 1

Observed mortality rate and expected mortality rate for myocardial infarction, based on adjusted ARIMA transfer function model, São Paulo city, 2005–2010. ARIMA, Autoregressive Integrated Moving Average.

The model estimated using ITSA–ARIMA for mortality rate from myocardial infarction was adjusted for total hospital admission, minimum air temperature and maximum CO level (table 4). This model showed a trend towards stability for mortality rate before the law, and an immediate decrease in level (compared with the counterfactual in the period immediately pos-intervention) (β2 −0.89; p<0.001). There was no trend towards change after the intervention (figure 2).

Table 4

Interrupted time-series analysis with ARIMA error model for mortality rate due to myocardial infarction, city of Sao Paulo, Brazil, January 2005 to December 2010

Figure 2

Observed mortality rate and expected mortality rate for myocardial infarction, based on adjusted ITSA–ARIMA transfer function model, São Paulo city, 2005–2010. ARIMA, Autoregressive Integrated Moving Average; ITSA, Interrupted Time Series Analysis.

The model estimated using ARIMAX for SUS hospital admission rate for myocardial infarction was adjusted to total hospital admission, air relative humidity, minimum air temperature and maximum CO level (table 5). This model showed an immediate effect of the smoking ban law (ω0t−0=−0.78, p=0.022), reducing monthly myocardial infarction rate in 5.4%. This effect persisted for 3 months after the smoking ban law (ω1t−3=0.83, p=0.036) (figure 3). According to this model, 142 myocardial infarction were avoided in the first 3 months after the ban.

Table 5

Transfer function model for SUS hospital admission rate for myocardial infarction, city of Sao Paulo, Brazil, January 2005 to December 2010

Figure 3

Observed rate and expected rate for hospital admission for myocardial infarction, based on adjusted ARIMA transfer function model, São Paulo city, 2005–2010. ARIMA, Autoregressive Integrated Moving Average.

The model estimated using ITSA–ARIMA for SUS hospital admission rates for myocardial infarction was adjusted for total hospital admission, air relative humidity, minimum air temperature and maximum CO levels (table 6). This model showed a trend towards a decrease in the hospital admission rate before the law. After the law, it was not observed any significant long-term effect in time series (figure 4).

Table 6

Interrupted time-series analysis with ARIMA error model for SUS hospital admission rate for myocardial infarction, city of Sao Paulo, Brazil, January 2005 to December 2010

Figure 4

Observed rate and expected rate for hospital admission for myocardial infarction, based on adjusted ITSA–ARIMA transfer function model, São Paulo city, 2005–2010. ARIMA, Autoregressive Integrated Moving Average; ITSA, Interrupted Time Series Analysis.

Discussion

In this study, a monthly decrease of almost 12% was observed in mortality rate for myocardial infarction in the first 17 months after the enactment of the comprehensive smoking ban law in São Paulo city.

Additionally, a monthly decrease of 5.4% was found in SUS hospital admissions for myocardial infarction during the first 3 months after the law. After this period, no additional reduction was observed and the hospital admission rate returned to the level before the law.

In comparison with previous studies, most of the evidence in the literature for rates of hospital admission is based on hospital admission data,21–25 whereas studies analysing population-based registries are limited.26–28 Concerning these previous data, most of the studies showed a decrease in hospital admission for acute myocardial infarction after smoking ban laws.

Regarding studies of cardiovascular mortality, only few studies examined the effect of smoking ban laws, and the results were divergent. Shetty did not find a significant decline in mortality for acute myocardial infarction in the USA when comparing areas of smoking restrictions with control regions, although there are some questions about the methodology concerning coverage of local smoking ban laws.29 Rodu studied the behaviour of myocardial infarction mortality in six states of the USA where smoking ban laws started and his conclusion was that local smoke-free ordinances had no immediate effect on myocardial infarction mortality.30

On the other hand, Villalbi showed a decrease in the mortality rate for acute myocardial infarction in Spain, in a population-based registry,31 and the magnitude of the reduction was similar to ours. Stallings-Smith studied an Irish population and showed a larger reduction in mortality rate for ischaemic heart disease.32 These two studies have a similarity to ours—they used a population-based registry. Villabi analysed population separated by sex and age. Stallings-Smith analysed other causes of mortality. In contrast, none of them analysed the environmental influence on these rates as we did.

In the present study, the mortality rate shown is representative of the entire population of the city (11 million people), and we found a mean decrease of 11.9% in the mortality rate during the 17 months after the law was enacted. Moreover, we analysed population data and environmental components together, looking for the influence of temperature, humidity and air components and pollutants in cardiovascular diseases. It is known that very high temperatures and very low temperatures increase the rate of cardiovascular events. Some pollutants, such as CO and particulate matter, have been previously described as causative of cardiovascular diseases.33–36 In this context, we considered it very important to analyse the influence of variations in these components in the behaviour of the rates of events, which was possible using ARIMAX models.

Concerning hospital admissions, most of the studies showed a reduction in hospital admission rates in the period of 1–2 years after the law was enacted. One of the reasons that could explain the fleetingness of the effect for hospital admissions is a peculiarity of the health system in São Paulo city: it is composed of the public health system (SUS), which is responsible for ∼55% of the medical coverage in the city, and a private health system, which is responsible for ∼45% of the medical care in the city. Moreover, Sao Paulo had a previous law with partial restriction to smoking in some closed places, like public repartitions and hospitals.

Then, we imagine that the population with the highest benefit after the comprehensive law is the patrons of bars, restaurants and other establishments not included in previous partial laws, including workers of these places. Most of this population are covered by private health system and is not included in this analysis.

The data described in this paper for hospital admission come from the public health system, and we believe the result could be more impactful if we had the entire population data, as we had for mortality rate. Concerning private hospitals, until 2010 there was not a standardised data acquisition system of hospital admissions for private health system and, because of this, it was not possible to acquire these data.

Considering pollutant data, some previous studies conducted in Sao Paulo and in other places, including a meta-analysis, showed a significant association among CO level and particulate matter with cardiovascular diseases.34 ,37 ,38 Regarding environmental variables, temperature variations are implicated with the occurrence of myocardial infarction.39 ,40 In our study, we used the pollutant and environmental data to estimate models using ARIMAX method and since the variations of air pollutants are related themselves, we opted for using that one which provided the best prediction of events and that has biological plausibility with cardiovascular diseases.

Changes in the treatment of myocardial infarction and stroke could interfere with the mortality and admission rates for these diseases. We looked for advances in medical care and did not find any change in the treatment of these diseases during the study period.

Moreover, we analysed the number of current smokers (%) during the same years studied for mortality and hospital admissions and did not find a significant decrease in the percentage of current smokers, which could influence the number of deaths occurred during the period included in our study. These data are from Surveillance for risk factors and protection against chronic diseases by telephone survey (VIGITEL) and are represented in table 7.42–47 There is an important source of tobacco information in Brazil, the Global Adult Tobacco Survey (GATS), which provided data about tobacco use, secondhand smoking and other important information to monitor the WHO Framework Convention on Tobacco Control.48 However, the survey was carried out in 2008 and 2013 and is programmed to be carried out again in 2018. Then, for this analysis, the period between the two surveys is too long.

Table 7

Active smoking population in Sao Paulo city (%), Brazil, according to VIGITEL—2006 to 201142–47

Analysing VIGITEL data, we observed that the influence of the smoking ban law in reducing current smoking is slower than for secondhand smoking, since only in 2012 and 2013 a reduction was observed in the percentage of current smokers (15.5% and 14.9%, respectively).49 ,50 In summary, the major immediate effect of the smoking ban law is in reducing secondhand smoking.

The two different statistical methods (ARIMAX and ITSA–ARIMA) used to estimate transfer function models showed immediate and consistent effects of the smoking ban law. Concerning mortality rate, it was shown an immediate and sustained decrease, of great magnitude (it is seen in the two methods).

Regarding hospital admission rate, the effect was shorter (3 months) and of smaller magnitude. Then, it was only saw in the ARIMAX model, in which it is possible to estimate two ω parameters (ω0t−0 e ω1t−3) and one of them had a time lag.41 On the other hand, ITSA aimed to detect immediate level change (before lagged) or long-term trend.

Finally, a comprehensive national smoking ban law was approved in Brazil in December 2014. It was one more important step in the Brazilian policy on tobacco control. Future researches will be necessary to analyse the impact of this law.

Limitations

This was a time-series study. Also, we analysed a population-based registry, not individuals separately. Therefore, we were not able to access the smoking status individually. In general, the rate of smokers varied from 18.8% to 21.7% in the years included in this analysis.42–47 Other individual information, such as cholesterol levels, diet, body mass index, family and personal history of coronary heart disease, exercise status, diabetes and hypertension were also not available.

Additionally, data about hospital admissions were not for the entire population of the city, because the private healthcare system in São Paulo is very complex, distributed among health insurance companies, cooperatives, self-management and group medicine. The public health system is responsible for ∼55% of medical care in the city.

Moreover, we are not able to separate the effects of the comprehensive smoking ban law and the effect of other antismoking policy changes, like increase in tobacco tax rates, implemented along the years, and especially in 2009.

Conclusion

Mortality rate and hospital admission rate for myocardial infarction decreased after the comprehensive smoking ban law in São Paulo city.

What this paper adds

  • Previous studies and meta-analyses demonstrated benefits of the smoking ban laws, reducing myocardial infarction rate. Some studies showed benefits in reducing myocardial infarction mortality rate too.

  • Most of the previous studies analysed only the ‘before and after’ the smoking ban law. We did not find other studies which analysed environmental components and pollutants in the same period of the law effect, in a large population as ours.

  • We analysed the ‘before and after’ the comprehensive smoking ban law in Sao Paulo city and analysed, simultaneously, environmental components and pollutants which have influence in the incidence of cardiovascular events.

References

View Abstract

Footnotes

  • Correction notice This article has been corrected since it was published Online First. The values in the final column of Table 2 have been updated. Data concerning monthly number of deaths for myocardial infarction was provided by Health Ministry through Mortality Information System and DATASUS. During the registration of information in table 2, the last column mistakenly recorded the sum of the number of stroke deaths with the number of stroke and infarction added, that is, unreal numbers. The correct numbers are now shown. The information can be obtained in http://tabnet.datasus.gov.br/cgi/tabcgi.exe?sim/cnv/obt10SP.def. This information will not interfere with the rest of the analysis, because the analysis was done will the original data of deaths, not based in this table.

  • Contributors TMOA conducted the study and submitted the article. JS planned the study and reviewed the article. EdM made the statistical analysis and help in the article writing. MRCN and RKF planned the study.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.