Ischemic heart disease (IHD) is one of the main causes of morbidity and mortality in the world, responsible for more than 7 million deaths annually. The mortality rate due to IHD began to increase from the beginning of the 20th century, but in most industrialized countries this trend has gone into reverse in the last 30 years.1,2 In contrast, in many middle- and low-income countries mortality due to IHD has continued to increase. This is due to the lack of access in these countries to diagnostic and therapeutic resources and poorer control of modifiable risk factors such as unhealthy eating habits, physical inactivity, overweight and obesity, smoking, diabetes mellitus, high blood pressure, and dyslipidemia.3

Historically, the effects of these modifiable risk factors has been studied in depth, but now the influence of the environment on cardiovascular health is gaining increasing attention in the context of global environmental change. The World Health Organization estimates that 24% of worldwide morbidity and 23% of mortality is attributable to environmental factors. Given that human beings interact continuously with the environment, variables such as air pollutants and meteorological conditions are potential predictors of cardiovascular morbidity and mortality.4,5

Biometeorological indices provide a measure of the combined effect of a set of environmental variables on human health and wellbeing. The biometeorological index apparent temperature (AT)6 estimates the human perception of temperature and has been identified by some authors as the main predictor of heat-related mortality.7 To date there have been no reports of the relationship between AT and cardiovascular morbidity.

The present study analyzed the relationship of AT and major air pollutants with hospital admissions for acute myocardial infarction (AMI) in Cantabria, a coastal region in northern Spain. The study also included analysis of patient demographic characteristics and associated mortality, as well as a temporal analysis to identify the disease seasonality and long-term trends in relation to findings from other countries.

The study was divided into 2 phases. The first phase was a descriptive study of AMI admissions and associated mortality at the Marqués de Valdecilla University Hospital (HUMV in its Spanish initials). The HUMV is the referral center in Cantabria and the only hospital in the region with an acute coronary unit. The second phase was a retrospective correlation analysis to determine the relationship between admissions for AMI and various environmental factors.

Apparent Temperature

AT is a biometeorological index that incorporates air temperature, humidity, and wind speed.

Mean AT is calculated from these meteorological variables using the formula proposed by R. Steadman (1984)8:

where T is the temperature in °C, Pv is the water vapor pressure (hPa), and V10 is the wind speed 10 m above the ground.

Water vapor pressure (Pv) was estimated with the following equation:

where rh is the percent relative humidity.

Statistical Analysis

Month-to-month and day-to-day differences were evaluated by analysis of variance. The annual incidence trend was calculated using the Kendall nonparametric Tau b test.

To evaluate the association between environmental variables and AMI admissions, we developed a distributed lag nonlinear model. In the time-series of pollutant data, 6.2% of values were missing. The missing values were concentrated in 2001 and 2002, prompting us to exclude these years from the analysis.

The distributed lag nonlinear model simultaneously describes complex nonlinear and delayed effects of an environmental variable on a response variable within generalized linear models, generalized additive models, or generalized estimating equations. The statistical details of the model have been described by Bhaskaran et al.9 and Gasparrini et al.10

The health effects of environmental variables are not always immediate, and lag periods have been reported of days or even weeks.11 To model the effect of AT and air pollutants we considered a 14-day lag period.12–15 A quasi-Poisson regression was enacted within a generalized additive model.

To evaluate the effect of AT, a first model was developed according to the following formula:

In this model, the potential confounding effects of pollutants were adjusted for by including the pollutant PM10, which was considered representative of the overall pollution level.

The second model was designed to evaluate the effect of each pollutant as an exposure variable according to the formula

where t is the observation day; Yt is the number of recorded admissions on day t; α is the intersection; Eatt;l, EPM10t;l, and Ect;l are the matrices obtained after applying the distributed lag nonlinear model to AT and the air pollutants; βi represents the coefficient vectors for these matrices; l is the lag in days; s(…) is a thin plate regression spline; trend is the long-term trend and seasonality on a natural cubic timescale with 7 degrees of freedom (df) per year; dow is the day of the week on day t, and η represents the coefficient vectors. The reference day of the week was Sunday.

Influenza is the number of cases of influenza, included with 3 df to smooth the effect of daily fluctuations, a possible confounder. For the rate of influenza infection, a locally weighted polynomial regression (LOESS) was used to extrapolate weekly data and obtain an estimate of daily data. We chose a14-day smoothing window, which corresponds to a behavior close to that of the reported cases.

The numbers of df for independent variables were selected according to the Akaike information criterion. The model assumed that the effect of exposure to the environmental variables was linear, and the lag effect was constructed using a 3 df cubic B-spline.

Model development, statistical analysis, and graph preparation were performed in R version 3.3. Models were analyzed with the software packages mgcv (version 1.8-15) and dlnm (version 2.2.6).

This is the first study to analyze the effect of the biometeorological index AT on the incidence of AMI in Europe. Given that AT is considered the best predictor of heat-related mortality,7 AT-related mortality data have strong potential as a prevention tool, and this forms the main focus of this discussion.

A downward trend in hospital admissions for AMI has been reported in several countries,16,17 and has generally been attributed to the success of primary prevention measures, which have produced decreases in smoking rates, systolic blood pressure, total cholesterol, and physical inactivity.18 This trend has been noted in several Spanish regions,19 and some authors have highlighted specific factors that may be important in the national context. One proposed factor is an increased uptake of a Mediterranean diet rich in virgin olive oil and nuts due to greater awareness of its health benefits for those at cardiovascular risk.20 Another likely important factor is the enactment of legislation on January 1, 2006 to restrict the sale, supply, consumption, and advertising of tobacco products (Ley 8/2005, de 26 de diciembre). The introduction of these restrictions has been associated with a decreased incidence of infarctions among active and passive smokers in Spain.21

The downward trend in mortality due to cerebrovascular disease and especially AMI in the United States, Canada, and western European countries has been attributed to the inclusion of troponins in diagnosis and the general improvement in treatment strategies. Moreover, some authors propose that the decrease in cardiovascular mortality reflects improvements in blood pressure control in recent years.22,23

The weekly pattern showed a higher admission rate from Monday to Friday; however, the most interesting result is the absence of significantly increased mortality at the weekend. Several studies have reported increased mortality among AMI patients admitted at the weekend, and this effect has been linked to lower resource availability and subsequent delays in effective treatment initiation.24 Supporting this idea, a recent UK study found no weekend mortality spike in hospitals with a specialized cardiology unit providing a full service 7 days a week.25

Other studies have reported peak AMI incidence in winter in a variety of climates.26,27 However, the available data are not fully consistent, with a higher incidence reported for the warmer months in some tropical regions, suggesting an influence of climate type.28

The study of admissions for AMI is considered a useful means of evaluating the impact of low temperatures on health.29,30 In these studies, it is important to recognize that a direct or inverse association of morbidity and mortality with temperature can be influenced by factors such as the geographic conditions of the country or region, specific population characteristics such as heat sensitivity, and adaptive measures.10,31 In the present study, we evaluated the effects of AT (rather than maximum, mean, or minimum temperature) because this measure attempts to capture the reaction of the human body to the combined conditions of the thermal environment. The human body is simultaneously exposed to a multitude of atmospheric parameters and maintains body temperature by regulating heat emission. Highly humid conditions impede evaporation and induce feelings of discomfort. Humidity is included as a variable in the calculation of AT.

A recent Portuguese study examined the association between AMI admissions and another biometeorological index, Physiologically Equivalent Temperature (PET).32,33 Like AT, PET incorporates temperature, humidity, and wind speed and provides a measure of the intensity of thermoregulation required to maintain bodily comfort. However, PET differs from AT in important respects; whereas AT is an empirical measure, PET is based on a model of bodily energy balance and incorporates the effect of solar radiation. Nonetheless, the measures are comparable, and the results of the Portuguese study are in agreement with our findings, showing an association between exposure to low temperatures and hospital admission for AMI.

These results support the controversial idea that low temperature exposure is prejudicial to human health even in countries with a mild cold season.34

An association between low temperatures and the incidence of AMI has also been reported for other European countries with climates less similar to Spain.35,36

The direct relationship between AMI admissions and PM10 is in line with findings in other locations. PM10 includes a proportion of anthropogenic materials (organic and secondary inorganic compounds) that are damaging to health upon short-term exposure. This material can be directly toxic to the circulatory system since the soluble particle components cross the alveolar epithelium; however, the deleterious effects usually act through indirect mechanisms involving the stimulation of oxidative stress (pulmonary and systemic) and inflammation. Moreover, vascular inflammation can activate a cascade of prothrombotic events, alter autonomic nervous system function (with effects on heart rate), and cause endothelial dysfunction.37 Air quality in Cantabria is good, with pollutants rarely exceeding World Health Organization limits,38 and this may explain the absence of a statistically significant relationship with other major air pollutants. However, this lack of association could also reflect certain study limitations. For example, the air pollutant data analyzed in this study came from a single monitoring station. This station was selected as the most appropriate for this type of study due to its general characteristics (urban and with a complete historical record) and specific location in the center of the largest city of the region. However, Cantabria contains a number of contrasting environments, including high traffic areas, industrial neighborhoods, rural and urban zones, and areas of low and high altitude. In these different areas, the pollutant concentrations to which people are exposed likely differ from those considered here as the most representative. The study is also limited by the small sample size, which reflects the small population of Cantabria and the correspondingly low number of AMI admissions.

It would have been interesting to analyze the effect of particulate matter with a smaller diameter (PM2.5). However, pollution monitoring stations in Santander did not begin recording this variable until 2009, and PM2.5 monitoring since then has not been continuous. It was therefore not possible to consider the effect of PM2.5.

The findings presented here indicate that environmental factors make an important contribution to mortality due to AMI. Further research directed at understanding these effects is essential for the development of prevention strategies.

None declared.