Association between Short-Term Exposure to Criteria Air Pollutants and Daily Mortality in Mexico City: A Time Series Study

Abstract

The short-term effects of air pollution on the health of residents in the metropolitan area of Mexico City (MAMC) were assessed in 11 municipalities from 2012 to 2015 using a time-series approach. Site 11 (Azcapotzalco) presented values above the limit of the Mexican regulations for SO₂, while values above the limit were found for CO for the city’s other municipalities. Site 8 (Cuauhtemoc) presented the highest number of values above the maximum permissible limit for NO₂, while site 1 (Alvaro Obregon) presented the highest number of values above the limit for O₃. Finally, site 7 (Venustiano Carranza) presented the highest number of values above the limit for PM₁₀. In general, the southeast and northwest of the city presented high levels of pollution for the criteria air pollutants: SO₂, NO₂, and PM₁₀, while the southeast presented the highest levels for O₃. A great number of associations were found between daily mortality and a 10% increase in the concentrations of most of the pollutants tested, for most of the municipalities of the city. Significant relative risk index (RRI) increases were found for people >60 years of age for all pollutants and municipalities, increases which resulted from a 10% increase in the daily mean concentrations of all pollutants tested. As the RRIs observed were low but significant, the findings are, thus, of public concern. The present study demonstrated that older people are at considerable risk from atmospheric pollution.

1. Introduction

Environmental pollution constitutes a health emergency affecting the populations of both low and high-income countries, as exposure to different environmental risk factors increases the occurrence of disease [1]. The Global Burden of Disease Study 2019 estimated that air pollution was responsible for 6.7 million deaths on a global level [1]. Due to the different nature of emission sources, atmospheric pollutants vary greatly in size, morphology, and chemical composition, and, depending on size and toxicity, some present adverse effects on human health. Therefore, air quality at a given site can be affected by diverse factors such as vehicular and industrial emissions, urban mobility, and the kind of fuels used. In addition, the presence of pollutants in the atmosphere, in either the short or long term, affects human health [2,3]. It has been reported that air pollution is the main risk factor for circulatory and respiratory diseases [4] and is, therefore, one of the main causes of death and disease in the world.

There is evidence that short-term exposure to air pollution is associated with respiratory and circulatory diseases and increased numbers of hospital admissions (morbidity) [5,6]. Among atmospheric pollutants, criteria air pollutants demand special attention, as they are present in ambient air at high concentrations from a great variety of sources, constituting a high risk to public health. The criteria air pollutants comprise sulfur dioxide (SO₂), nitrogen dioxide (NO₂), carbon monoxide (CO), particulate matter (PM), and ozone (O₃). The health effects associated with these pollutants are mainly acute and non-carcinogenic. In terms of acute effects, it has been documented that increased PM₁₀ and O₃ levels in the atmosphere correlate well with early mortality [7,8,9,10,11].

In recent years, epidemiological studies have determined the association between exposure to environmental risk factors and health effects, considering daily mortality by all causes and specific cause (circulatory and respiratory), as well as by gender and age group. In this regard, some epidemiological studies have reported that respiratory diseases that develop in adult populations have their origin in early stages of life [12,13,14]. In addition, the association between exposure to air pollutants and respiratory diseases (pneumonia, asthma, and acute respiratory infections) in vulnerable populations (children) has been documented by many authors [15,16,17,18,19]. These studies report that, the exposure of children to environmental pollutants leads to an increased risk of chronic respiratory diseases, such as chronic obstructive pulmonary disease, asthma, accelerated or premature lung aging, and, therefore, reduced life expectancy [20,21,22]. Short-term exposure to air pollutants such as PM₁₀ (particulate matter with a diameter <10 μm) and nitrogen dioxide (NO₂) has also been associated with mortality, and respiratory, circulatory, and metabolic diseases [2,3,23,24]. Yung [25] reported that CO, SO₂, O₃, and PM₁₀ exposure are highly associated with mortality, while it has also been documented that short-term SO₂, NO₂, and O₃ exposure are associated with certain effects on the indicators of metabolic health in patients with metabolic syndrome [26].

In Mexico, some epidemiological studies have assessed the association between atmospheric pollutants and mortality; however, there are not enough data or available information to describe the current prevailing situation in most metropolitan areas of the country. Therefore, the present study aimed to estimate the association between criteria air pollutants and daily mortality from 2012 to 2015 in 11 municipalities within the MAMC, considering an analysis by gender, age group, and specific cause of death (respiratory and circulatory diseases). In addition, to estimate the association between air pollutants and mortality, the present study assessed the effects of a hypothetical scenario in which the daily mean concentrations of atmospheric pollutants increases by 10%. This hypothetical scenario enabled the quantification of the number of concentration values above the limit of the regulations for each pollutant in each municipality, and, finally, the identification of the municipalities in which the relative risk was higher and the population sub-group of interest more vulnerable.

2. Materials and Methods

2.1. Study Area

The MAMC is one of the world’s largest metropolitan areas and is the largest city in North America, with a population of approximately nine million inhabitants. Because Mexico City is located in a valley, its topography and meteorological conditions are complex and do not promote the effective dispersion of air pollutants. Therefore, both a notably large population and poor air quality coexist in this area. Although Mexico City is divided into 16 municipalities, solely the 11 municipalities for which validated air quality data were available were included in the present study. The location of the municipalities is shown in Figure 1.

2.2. Air Quality and Meteorological Data

Air quality pollutants levels (CO, NO₂, SO₂, O₃, and PM₁₀) and meteorological data (temperature and relative humidity) for 11 municipalities in the MAMC were obtained from the National Air Quality Information System (SINAICA: Sistema Nacional de Información de la Calidad del Aire) from 2012 to 2015. Criteria air pollutants were measured via an automatic monitoring network using standardized methods (

Table 1. Standardized methods used for the measurement of each criteria air pollutant.

Criteria Air Pollutant	Principle of Operation	Method Description
Sulfur dioxide (SO2)	UV fluorescence	Measurement of the fluorescence emitted when SO2 molecules are excited by an UV radiation source
Carbon monoxide (CO)	Infrared absorption	Measurement of the absorption of infrared light by carbon monoxide in a correlation cell
Nitrogen dioxide (NO2)	Chemiluminescence	Measurement of the light emitted during the reaction between NO and O3. The separation of nitrogen species is carried out via the differential measurement of NO and NO2 (with a previous catalytic reduction). The corresponding value for NOx is equal to NO + NO2.
Ozone (O3)	Photometry UV	Measurement of UV light absorption at a wavelength of 254 nm. The reduction in the signal intensity is proportional to ozone concentration according to the Beer–Lambert law.
Particulate matter (PM10)	Gravimetric	Measurement of the mass of particles in the airflow. The particles separate from the stream and are then deposited on a filter placed on an oscillating element. The variation of the oscillation frequency is proportional to the mass. The particle size is determined by both selective entry and the sampling flow.

).

Meteorological parameters related to health impact were included in the analysis as they are explicative variables for the association between atmospheric pollutants and daily mortality. Air pollution increases during extremely cold and extremely warm periods due to the intensive use of heating and air conditioning equipment, which results in higher emissions of atmospheric pollutants. In addition, during the winter season, thermal inversions that inhibit the dispersion of pollutants are frequent in Mexico City [28]. Moreover, the probability of deaths in the vulnerable population (children under five years old and adults older than 60 years) increases with extreme temperatures during the warm and cold months, while relative humidity has a negative impact on health by changing the conditions in which some diseases develop.

2.3. Applicable Regulatory Framework

The Mexican Federal Government is the responsible for establishing reference values in order to protect public health. These values are published as the Official Mexican Standards (NOMs) and are obligatory at a national level. The maximum permissible levels established in the NOMs are as follows:

• For O₃: 0.090 ppm (1 h average data) and 0.065 ppm (8 h moving average data) [29];
• For CO: 26.0 ppm (1 h average data) and 9.0 ppm (8 h moving average data) [30];
• For NO₂: 0.106 ppm (1 h average data) and 0.021 ppm (annual average) [31];
• For PM₁₀: 70 µg m⁻³ (24 h daily average data) and 36 µg m⁻³ (annual average) [32]; and,
• For SO₂: 0.075 ppm (1 h average data) and 0.040 ppm (24 h daily average data) [33].

The present study estimated the values above the reference values established for each criteria air pollutant, while the time series for each pollutant and each meteorological variable were compiled considering average daily and maximum daily values. In some cases, data obtained from automatic monitoring stations do not always enable continuous registering, while, in other cases, data for a specific pollutant are missing. Therefore, to complete the database for the 2012–2015 period for each station, it was necessary to establish some inclusion criteria, firstly, to decide which stations would be included and, secondly, to establish whether data were missing from one or more of the stations. Therefore, it was necessary to define how missing data would be imputed to complete the database. The following criteria were applied:

Valid data percentage: The estimation of mean concentrations for each pollutant solely considered values from monitoring stations that presented a >75% valid data percentage. In addition, it was necessary to apply an imputation method for stations consistently presenting missing data despite providing valid data percentages.

Imputation of missing data: Isolated or intermittently missing data from a database corresponding to a monitoring station were completed using the NIPALS approach to imputing missing data.

2.4. Epidemiological Data

Studies on air pollution exposure and its effects on health are scarce in developing countries. The main challenge in this field is to collect reliable data in the fields of both health and air quality, mainly because the collection, storage, and validation procedures are neither continuous nor uniform in these countries. While SINAICA makes great efforts to collect air quality information for several cities in Mexico, there is no national system for collecting health data (mortality and morbidity). For this reason, the present study obtained mortality data for each municipality in Mexico City on a monthly and annual basis from the database maintained by the National Institute of Statistics and Geography (INEGI). The INEGI mortality database uses the international classification of diseases established by the World Health Organization (WHO) review CIE-10/2 for respiratory system (J00–J99) and circulatory system diseases (I00–I99). Therefore, the mortality data were assessed by cause of death (all causes, respiratory, and circulatory), gender (male and female), and age group (<1 year, 1–4 years, 5–59 years, 60–74 years, and >75 years).

2.5. Design of the Epidemiological Analysis

The design of the epidemiological analysis adhered to the following stages:

• Description and assessment of temporal variations in mortality rate on a monthly basis for all populations and all population sub-groups, by age, gender, and specific cause of death from 1 January 2012 to 31 December 2015.
• Description and assessment of temporal variations in criteria air pollutants concentrations on a monthly basis from 1 January 2012 to 31 December 2015.
• Diagnosis of air quality: Values above the reference values established as maximum permissible limits in the NOMs for each criteria air pollutant and each municipality were obtained.
• Estimation of the magnitude of the association between mortality (by all causes of death and by specific cause of death) and air pollutants concentrations for each municipality and population sub-group. During this stage, the meteorological variables (temperature and relative humidity) were included.

Study Subject:

Number of deaths occurring in the study area from 1 January 2012 to 31 December 2015.

Variables:

(A)

Response variables: Number of monthly and daily deaths for each municipality during the study period, considering different causes of death: respiratory, circulatory, and all causes.

(B)

Explanatory variables: Criteria air pollutants (quantitative explanatory variable), monthly and daily mean concentration values for O₃, SO₂, CO, NO₂, and PM₁₀ for each municipality during the study period. Meteorological variables (quantitative explanatory variable): monthly and daily mean values for temperature and relative humidity for each municipality during the study period. Gender (qualitative explanatory variable): number of deaths by gender. Age (qualitative explanatory variable): number of deaths by age group.

(C)

Control variables (seasonality): Seasons were classified as cold months (from November to February) and warm months (from May to August).

(D)

Confusion variables: Temperature and relative humidity

(E)

The delay in the effect of the confusion variables was considered: the time series were pretreated, wherein the time delay from cross correlations of the series (mortality vs. temperature and mortality vs. relative humidity) was estimated by using Infostat software v. 2008 [34], with time delays selected according to their significance level. The epidemiological data series were smoothed, as they mostly presented collinearity and non-linear relationships with some variables (for example, temperature and relative humidity); therefore, a non-parametric method (LOWESS: locally weighted regression scatterplot smoothing) was applied. The air quality data series were pretreated (smoothed) using the ARIMA method (autoregressive integrate moving average) (Box–Jenkins model) [35].

2.6. Estimation of the Magnitude of the Association between Mortality, by All Causes of Death and by Specific Cause of Death, and Atmospheric Pollution Concentrations for Each Municipality and Population Sub-Group

Once the time series had been treated and smoothed, it was possible to apply a Poisson model, thus reducing Pearson residuals. The time series were smoothed using XLSTAT statistical software v. 2017 “https://www.xlstat.com/es/ (accessed on 17 September 2022)”.

2.6.1. Multivariate Analysis

A principal component analysis (PCA) and a linear regression analysis (LRA) were applied to the data series for mortality, criteria air pollutants, and meteorological variables. The PCA was used to obtain the principal components contributing to the highest percentage of data variability (F₁ and F₂ axes) [36,37]. The PCA and LRA were carried out using XLSTAT software v. 2017. To confirm whether a group of variables were strongly correlated, the cosine squared table was assessed. Each factor included a set of variables, which represented the degree of association among them, taking into account those presenting the highest factor loadings and greatest statistical significance. The criterion used to ascertain whether or not to include a variable in the basal model was a p < 0.10 deviance. A first approach was attempted with the basal model including all variables by cause of death and pollutant. Based on the PCA and LRA analysis, the variables that contributed largely to explaining the variability of the data of the dependent variable (daily mortality) were identified and then they were included in the Poisson model. The foregoing was carried out considering different causes of death: respiratory, circulatory, and all causes of death, different age groups, and stratified analysis (warm and cold months).

2.6.2. Relative Risk Index for Daily Mortality Associated with Atmospheric Pollution

For each mortality series, a Poisson regression model was built to explain the fluctuations in daily mortality relative to explanatory and confusion variables. The same methodology as APHEA [38] and EMECAN [39] was used for the application of the Poisson model. First, a basal model was identified for each cause of death from the possible confusion variables. Once the basal model had been established, the model was extended for each pollutant and its time delays. The construction of the auto-regressive Poisson model allowed determining if the response variable depends, or not, on other variables. If independent variables have a significant effect on the response variable (for a confidence interval of 95% and p < 0.05), these effects can be assessed by means of the beta coefficient of each independent variable in the Poisson regression model. The general model to relate the response variable with different independent variables was obtained as follows:

$$\ln \left(E_y\right)={\beta }_0+{\displaystyle \sum }_{i=1}^n{\beta }_i{x}_{t,i}$$

(1)

where E_y is the expected number of cases, ${\beta }_0$, ${\beta }_i$ are the model constants, and $x_{t,i}$ are the explanatory variables. Once the Poisson regression had been carried out using the basal model, the beta coefficients were obtained and used to estimate the RRI as follows:

$$RR{I}_i=e^{{\beta }_i}$$

(2)

where, RRI_i is the relative risk index associated to the explanatory variable i by increment unit of this variable, and ${\beta }_i$ is the regression coefficient associated to the explanatory variable i in the model. Subsequently, the concentration of each pollutant and meteorological variable considered by the model was increased by 10% separately. From this assumption, the regression parameters were obtained again and the Poisson distribution was then applied (taking this increase into account and keeping the remaining variables unchanged). With the obtained ${\beta }_i$ values from the fitted model, the relative risk index was estimated for mortality considering an increase of 10% in the magnitude of each explanatory variable in order to determine the derived effect from an increase in concentration for a given criteria air pollutant. This procedure was applied for each municipality and for each criteria air pollutant. The Poisson regression analysis was carried out using XLSTAT statistical software v. 2017 “https://www.xlstat.com/es/ (accessed 17 September 2022)”.

2.7. Mapping the Relative Risk Index for Each Municipality

Mapping of relative risk indexes (RRI) of daily mortality by all causes associated to each pollutant was carried out by using QGIS v. 2.14.7, a geographic information system, from spatial data of vector type including the geographically referenced municipalities. The information was obtained from the National Institute of Statistic, Geography and Informatics (INEGI).

3. Results and Discussion

3.1. Air Quality

Descriptive statistical information for each criteria air pollutant by municipality in Mexico City is shown in Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6. Figure 2 shows that SO₂ concentrations in the study period were uniform except for sites 7, 8, 9, and 11, wherein site 11 (Azcapotzalco) was the municipality which presented the highest concentrations for this pollutant. Sites 7, 8, 9, and 11 are located in the central and northern areas of Mexico City. Many industrial sources of pollution are located in these areas, which are characterized by commonly present high vehicular traffic flow.

Figure 3 shows that CO concentrations along the study period presented a high variability, with sites 4, 7, 9, and 11 being the municipalities that showed the highest concentrations values. Site 7 (Venustiano Carranza) presented the highest concentrations. Site 4 (Iztapalapa) is located in southeast Mexico City, whereas sites 7, 9, and 11 are located in the center and northern Mexico City.

Figure 4 shows a high level of variability in NO₂ levels during the study period. The highest concentrations were found at sites 7, 8, 9, 10, and 11. Site 7 (Venustiano Carranza), located in the north of the city, was the municipality which presented the highest concentrations. Sites 7, 8, 9, 10, and 11 are located in the center and north of Mexico City.

Figure 5 shows that O₃ presented the highest concentration levels at sites 1, 3, 5, and 6, located in the center, southwest, south, and southeast of Mexico City, respectively. The highest concentrations were found at site 3 (Cuajimalpa), located in the southwest of the city.

Figure 6 shows that the highest PM₁₀ concentrations were found at sites 4, 7, 8, and 11, located to the southeast, north, center, and north of Mexico City, respectively. Site 7 (Venustiano Carranza) presented the highest concentrations.

3.2. Epidemiological Data

The municipalities that presented the highest mean daily mortality values by all causes of death during the study period were Gustavo A. Madero and Iztapalapa, while Tlahuac and Xochimilco registered the lowest mean values (Figure 7). In terms of daily mortality, the standardized mortality rate was estimated as a function of the number of daily deaths per 1000 inhabitants in an assigned area and fitted by age distribution. The municipalities which presented the highest mortality rates were Cuauhtemoc, Venustiano Carranza, and Azcapotzalco, while Cuajimalpa and Tlahuac presented the lowest rates. It should be noted that, in spite of the fact that Azcapotzalco did not present a high number of daily deaths, it did present a high mortality rate, in this case, the second highest in Mexico City. A certain uniformity was observed in the results for each municipality, with the maximum values found during January, thus indicating a seasonal pattern, wherein the number of deaths was higher during the cold months and winter season.

In terms of age group, Figure 7 shows that, considering frequency distribution, mean diagrams, and descriptive statistics applied to daily mortality data by all causes, higher mortality values were found for the population >60 years, suggesting that this sub-group is the most vulnerable. On the other hand, children (aged between 0 and 4 years) were the least vulnerable subgroup. This finding was uniform and consistent in all municipalities studied. In all municipalities, women showed a higher number of deaths than men. The maximum numbers of respiratory causes of death were observed during January for adults >60 years, whereas the minimum numbers were found during the summer and warm months for children. Women registered the highest number of deaths for this specific cause of death. The maximum number of deaths due to circulatory causes were observed during January for adults >60 years, whereas the minimum values were found during the summer months for children. Women registered the highest number of deaths for this specific cause of death. Finally, women >60 years presented higher daily mortality due to circulatory causes than due to respiratory causes.

3.3. Values above the Reference Values Established by the NOMs

Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12 compile the data on the values above the limit for each pollutant and municipality. Figure 8 shows that sites 11 (Azcapotzalco), 9 (Iztacalco), and 7 (Venustiano Carranza), which are located in the north and center of the city, presented the highest mean SO₂ concentrations during the study period. However, site 11 (Azcapotzalco) presented the highest number of values above the maximum allowable limit established by the Mexican standard [33]. All sites exceeded the established air quality standards for SO₂ by both the World Health Organization (WHO) [40] and the United Kingdom (National Air Quality Standards Regulations) [41].

Figure 9 shows that sites 7 (Venustiano Carranza), 8 (Cuauhtemoc), and 9 (Iztacalco) were the municipalities with the highest CO concentrations. No municipality presented values above the maximum permissible limit established by the Mexican standards [30]. However, it can be observed that most of the study sites exceeded the values established as the air quality standard for CO by the WHO [40], whereas no site exceeded the limits established by the U.K. (National Air Quality Standards Regulations) [41] and U.S. standards (Environmental Protection Agency or EPA) [42].

Figure 10 shows that sites 7 (Venustiano Carranza), 8 (Cuauhtemoc), 9 (Iztacalco), and 11 (Azcapotzalco) presented the highest number of values above the maximum permissible limit established by the Mexican standards [31]. Except for sites 1 (Alvaro Obregon) and 6 (Tlahuac), all sites exceeded the value established by the NOM [31]. It can be observed that, except for site 6 (Tlahuac), that all sites exceeded the values established as the air quality standard for NO₂ by both the U.S. (EPA) [42] and U.K. (National Air Quality Standards Regulations) [41].

Figure 11 shows that no municipality exceeded the maximum allowable limit established in the Mexican standards [29]. It can be observed that no site exceeded the values established as the air quality standard for O₃ by both the U.S. (EPA) [42] and WHO [40].

Figure 12 shows that, except for site 3 (Cuajimalpa), all sites exceeded the maximum allowable limit established by the Mexican standards [32]. It can be observed that all sites exceeded the value established as the air quality standard for PM₁₀ by the WHO [40], whereas site 3 was the only municipality which did not exceed the air quality limit established by the U.K. (National Air Quality Standards Regulations) [41].

3.4. Estimation and Mapping of the Relative Risk Index for Each Pollutant by Municipality

3.4.1. Bi-Variate Analysis, Multivariate Analysis, and Multiple Regression of Daily Mortality Data with Explanatory Variables

High determination coefficients (R²) were found for daily mortality associated with respiratory diseases (which are more evident during the cold months in the >60 years age group). Significant inverse correlations were found among daily mortality, temperature, and relative humidity. The highest R² values (RLM) were found at sites 1 (R² = 0.9123), 10 (R² = 0.8415), 4 (R² = 0.8123), and 7 (R² = 0.7814), for all causes of death for the population aged over 60 years. A higher correlation was found at site 11 between mortality and all atmospheric pollutants tested for all age groups (0–59 years, R² = 0.8020; older than 60 years, R² = 0.8512), at site 10 for SO₂ (respiratory diseases, R² = 0.8134; and, circulatory diseases, R² = 0.7952), CO (during the winter season and the cold months, R² = 0.8510), and NO₂ (during the winter season and the cold months, for respiratory diseases, R² = 0.9044). A higher correlation was found at site 4 for PM₁₀ for respiratory diseases for all age groups (0–59 years, R² = 0.7714; and, >60 years, R² = 0.8936), at site 9 for PM₁₀ (for the age sub-group of >60 years, R² = 0.9027; and, for respiratory diseases, R² = 0.7804), and at site 1 (for >60 years, R² = 0.8917). The PCA analysis showed that the CO and NO₂ concentrations had a significant correlation with daily mortality at site 10 for all causes of death (R² = 0.7063) and for all age sub-groups, for respiratory (R² = 0.8345) and circulatory diseases (R² = 0.7135), during the cold months. The concentrations of O₃ presented associations with daily mortality at site 11 (for all causes of death, R² = 0.7926, for all age sub-groups, and for all circulatory diseases, R² = 0.8842) and at site 9 (for the 0–59 years sub-group, R² = 0.6825, for circulatory, R² = 0.7822, and respiratory diseases, R² = 0.7459). The concentrations of SO₂ presented a significant correlation with daily deaths at site 11 (R² = 0.8840) during the cold months, at site 8 (for respiratory diseases, R² = 0.7769), at site 7 (for circulatory diseases, R² = 0.8578), and at site 4 (R² = 0.8816, during the cold months).

3.4.2. Relative Risk Index Estimation

The present study found a significant association between daily mortality and CO (0.9996–1.0004), O₃ (0.9861–1.0091), and PM₁₀ (0.9904–1.0141) for circulatory diseases. Significant associations were found between daily mortality and atmospheric pollutants at site 11 (Azcapotzalco), site 8 (Cuauhtemoc), site 7 (Venustiano Carranza), site 1 (Alvaro Obregon), site 10 (Gustavo A. Madero), site 9 (Iztacalco), and site 6 (Tlahuac). Site 11 (Azcapotzalco) presented a significant association between daily mortality and NO₂ (0.9863–1.0151), O₃ (0.9875–1.0091), and PM₁₀ (0.9916–1.0149) concentrations for people >60 years, suggesting that this sub-group of the population in this municipality is highly vulnerable. High RRI values were found for PM₁₀ (0.9917–1.0135) when all causes of death were considered. Site 8 (Cuauhtemoc) presented the highest RRI values for CO (0.9993–1.0011), O₃ (0.9875–1.0212), and PM₁₀ (0.9789–1.0294) for the 0–59 years sub-group, suggesting that this age range is more vulnerable to the effects of atmospheric pollution in this municipality. An important association between daily mortality and CO (0.9996–1.0004) and NO₂ (0.9862–1.0127) was found for circulatory diseases. Site 7 (Venustiano Carranza) presented significant associations, for respiratory diseases, between daily mortality and SO₂ (0.9551–1.0389), CO (0.9992–1.0007), O₃ (0.9702–1.0191), and PM₁₀ (0.9844–1.0250) concentrations. Considering all causes of death, this association was only significant for O₃ (0.9889–1.0099). At site 1 (Alvaro Obregon), it was found that people >60 years are more vulnerable, with significant associations observed between mortality and SO₂ (0.9725–1.0258), CO (0.9995–1.0007), and NO₂ (0.9808–1.0149) levels. In addition, a higher level of association was observed between daily deaths and SO₂ (0.9716–1.0380) and NO₂ (0.9839–1.02649) concentrations for the 0–59 years sub-group and between daily deaths and SO₂ (0.9854–1.0190) for circulatory diseases. The concentrations of NO₂ presented a significant association with daily mortality for all causes of death and for the >60 years group (0.9880–1.0168) at site 9 (Iztacalco) (0.9903–1.0169), whereas, at site 6 (Tlahuac), this association was significant for the 0–59 years sub-group (0.9707–1.0238) and respiratory diseases (0.9762–1.0261). The association was significant during the cold months for CO, O₃, and PM₁₀ at site 8 (Cuauhtemoc), for SO₂ and CO at site 10 (Gustavo A. Madero), and for SO₂ and NO₂ at site 7 (Venustiano Carranza). During the warm months, the association was significant for SO₂ at site 1 (Alvaro Obregon), for PM₁₀ at site 7 (Venustiano Carranza), and for NO₂ at site 6 (Tlahuac).

3.4.3. Integrated Mapping of Relative Risk Index for Each Pollutant

Figure 13, Figure 14, Figure 15, Figure 16 and Figure 17 show maps detailing the RRIs generated for the 2012–2015 period for the hypothetical scenario in which SO₂, CO, NO₂, O₃, and PM₁₀ concentrations increased by 10% in each municipality of Mexico City. For all municipalities, a 10% increase in SO₂ levels would produce 0.22%, 0.18%, 0.04%, and 0.03% increases, respectively, in daily mortality at site 10 (Gustavo A. Madero), site 11 (Azcapotzalco), site 8 (Cuauhtemoc), and site 4 (Iztapalapa), respectively (see Figure 13).

The highest increases in daily mortality associated with a 10% increase in daily CO concentrations were: 0.007%, 0.006%, 0.006%, 0.005%, 0.005%, 0.002%, and 0.002% at site 7 (Venustiano Carranza), site 8 (Cuauhtemoc), site 10 (Gustavo A. Madero), site 11 (Azcapotzalco), site 1 (Alvaro Obregon), site 4 (Iztapalapa), and site 9 (Iztacalco), respectively (Figure 14).

The highest increase in mortality associated with NO₂ concentration levels increased by 10% was found at site 9 (Iztacalco), site 2 (Coyoacan), site 10 (Gustavo A. Madero), site 4 (Iztapalapa), and site 3 (Cuajimalpa), with values of 0.35%, 0.09%, 0.06%, 0.04%, and 0.02%, respectively (Figure 15).

A 10% increase in daily O₃ concentrations did not represent a significant risk of mortality in the municipalities studied in Mexico City (Figure 16).

A 10% increase in daily mean PM₁₀ concentrations corresponded to the highest increases in daily mortality being projected for site 11 (Azcapotzalco) (0.26%), site 4 (Iztapalapa) (0.13%), site 9 (Iztacalco) (0.09%), site 1 (Alvaro Obregon) (0.08%), site 7 (Venustiano Carranza) (0.04%), and site 3 (Cuajimalpa) (0.02%) (see Figure 17).

3.5. Comparisons with Other Studies

Table 2. Comparison of the RRIs and 95% CI obtained for SO₂ by the present study with other studies conducted around the world.

Site	>60 Years	Respiratory Cause of Death	Circulatory Cause of Death	All Causes of Death
Barcelona, Spain [43]	1.065–1.209	1.063–1.232	0.991–1.283	-
Lyon, France [44]	-	1.041–1.144	1.01–1.07	-
13 cities in Spain [45]	1.001–1.01	-	-	-
Zaragoza, Spain [46]	1.008–1.058	1.006–1.187	1.001–1.287	-
Sao Paulo, Brazil [47]	-	0.981–1.101	1.01–1.08	-
North Korea [48]	-	-	1.038–1.044	-
Chongqing, China [49]	1.00–1.09	-	-	-
China [50]	-	-	-	1.0125–1.0411
Various cities [51]	-	1.0025–1.0109	-	1.0046–1.0071
Site 1	-	0.9794–1.0204	0.9871–1.0083	0.9745–1.0233
Site 11	-	-	0.9956–1.0084	0.9962–1.0074
Site 2	0.987–1.0081	-	-	-
Site 3	-	0.9606–1.0071	-	-
Site 8	0.9836–1.0153	0.9746–1.0315	0.9838–1.0179	0.9858–1.0153
Site 10	0.986–1.019	0.9815–1.0312	0.9854–1.019	0.9865–1.0181
Site 9	0.9788–1.0216	-	-	0.9798–1.0197
Site 4	0.9812–1.0192	0.9909–1.0132	0.9809–1.0195	0.9826–1.0184
Site 7	0.9884–1.0039	0.9551–1.0389	0.9876–1.0049	0.9784–1.0154

,

Table 3. Comparison of the RRIs and 95% CI obtained for CO by the present study with other studies conducted around the world.

Site	>60 Years	Respiratory Cause of Death	Circulatory Cause of Death	All Causes of Death
13 Spanish cities [45]	1.005–1.026	1.014–1.051	1.013–1.032	-
Sao Paulo, Brazil [47]	-	0.981–1.101	1.01–1.08	-
Cities in Italy [52]	-	1.015–1.053	1.012–1.024	-
Cities in Europe [53]	1.009–1.018	1.008–1.03	-	-
272 cities in China [54]	1.0042–1.0183	-	1.0085–1.0266	-
Monterrey, Mexico [55]	0.9999–1.0006	0.9999–1.0003	0.9999–1.0002	-
Site 1	0.9995–1.0007	0.9996–1.0005	0.9999–1.0004	0.9995–1.0006
Site 11	0.9996–1.0005	0.9998–1.0004	0.9996–1.0004	0.9996–1.0005
Site 2	0.9997–1.0001	0.9996–1.0002	0.9998–1.0002	-
Site 8	0.9997–1.0005	0.9995–1.0009	0.9996–1.0004	0.9997–1.0004
Site 10	0.9998–1.0003	0.9996–1.0005	0.9998–1.0004	0.9998–1.0003
Site 9	0.9997–1.0004	-	0.9999–1.0002	0.9997–1.0004
Site 4	0.9997–1.0003	0.9998–1.0002	0.9997–1.0003	0.9997–1.0003
Site 7	0.9997–1.0004	0.9992–1.0007	1.0000–1.0003	0.9997–1.0004

,

Table 4. Comparison of RRIs and 95% CI obtained for NO₂ by the present study with other studies conducted around the world.

Site	>60 Years	Respiratory Cause of Death	Circulatory Cause of Death	All Causes of Death
Monterrey, Mexico [55]	-	0.9851–1.0122	0.991–1.0082	-
North Korea [48]	-	-	1.057–1.063	-
Cities in Canada [56]	1.045–1.059	1.053–1.083	0.977–1.034	-
272 cities in China [54]	1.007–1.0115	1.009–1.015	1.007–1.012	-
Panama City [57]	1.188–1.665	1.019–1.213	1.001–1.082	-
Italy [58]	-	1.0102–1.1113	1.0103–1.1395	-
China [59]	-	-	-	1.014–1.133
China [60]	-	-	-	1.042–1.219
DUELS project [61]	-	1.01–1.03	-	1.02–1.04
France [62]	-	-	-	1.0000–1.1500
London, U.K. [63]	-	-	-	0.76–1.17
Denmark [64]	-	-	-	1.04–1.17
Tehran, Iran [65]	-	-	-	0.999–1.007
Site 1	0.9808–1.0149	0.9887–1.0132	0.9885–1.0024	0.9824–1.0143
Site 11	0.9863–1.0151	-	-	0.9879–1.0147
Site 2	0.9994–1.0092	0.9948–1.0132	0.9958–1.0074	0.9928–1.0091
Site 3	-	-	0.9921–1.0117	0.9907–1.0099
Site 8	0.9854–1.0107	0.9691–1.0135	0.9862–1.0127	0.9865–1.0099
Site 10	0.9904–1.0106	0.9913–1.023	0.9894–1.0103	0.991–1.0104
Site 9	0.988–1.0168	-	-	0.9903–1.0169

,

Table 5. Comparison of the RRIs and the 95% CI obtained for O₃ by the present study with other studies conducted around the world.

Site	>60 Years	Respiratory Cause of Death	Circulatory Cause of Death	All Causes of Death
Monterrey, Mexico [55]	-	0.9849–1.0127	0.9912–1.008	-
Stockholm, Sweden [66]	1.01–1.03	-	-	-
Cities in Canada [56]	1.026–1.036	1.053–1.083	0.977–1.034	-
Panama City [57]	1.001–1.024	1.003–1.3	1.002–1.144	-
Cities in Italy [58]	-	1.00–1.09	1.01–1.06	-
England [67]	-	-	-	0.84–1.06
South London [63]	-	-	-	0.73–1.85
Hefei, China [59]	-	-	-	1.004–1.051
China [68]	-	-	1.04–1.17	1.08–1.17
China [60]	-	-	-	1.006–1.034
Europe [69]	-	-	-	1.0009–1.0025
Shanghai, China [70]	-	-	-	1.004–1.0058
Various cities [51]	-	-	-	1.0034–1.0052
Canada [71]	-	-	1.0002–1.011	-
Site 1	0.989–1.0068	0.9926–1.0049	0.994–1.0009	0.99–1.0062
Site 11	0.9875–1.0091	0.991–1.0069	0.9861–1.0091	0.988–1.0081
Site 2	-	-	0.9973–1.0037	-
Site 3	-	-	0.993–1.0078	0.9927–1.007
Site 8	0.9897–1.0061	0.9846–1.0139	0.9891–1.0062	0.9909–1.0061
Site 10	0.9931–1.005	0.9898–1.0092	0.9925–1.0053	0.993–1.0052
Site 9	0.9867–1.0068	0.9906–1.0048	0.9935–1.0013	0.9876–1.0061

and

Table 6. Comparison of RRIs and 95% CI obtained for PM₁₀ by the present study with other studies conducted around the world.

Site	>60 Years	Respiratory Cause of Death	Circulatory Cause of Death	All Causes of Death
Monterrey, Mexico [55]	-	-	0.9973–1.0027	-
North Korea [48]	1.009–1.013	-	1.136–1.149	-
Panama City [57]	1.071–1.58	1.002–1.242	1.058–1.136	-
Stockholm, Sweden [66]	1.008–1.014	-	-	-
Cities in Canada [56]	1.029–1.041	-	-	-
Pudahuel, Chile [72]	-	-	1.0007–1.0165	-
Temuco, Chile [72]	-	-	1.0004–1.025	-
Tehran, Iran [73]	1.004–1.008	-	-	-
Kermonshah, Iran [74]	-	1.005–1.022	1.005–1.021	1.004–1.008
Italy [58]	-	1.0128–1.0392	1.0006–1.022	-
Various cities [51]	-	-	-	1.0034–1.0049
Site 1	0.9911–1.0103	0.9926–1.0065	0.9973–1.0051	0.9919–1.0098
Site 11	0.9916–1.0149	0.9967–1.0131	0.9904–1.0141	0.9917–1.0135
Site 3	-	-	0.9957–1.0665	0.9943–1.0061
Site 8	0.99–1.0102	0.9831–1.0184	0.9896–1.0107	0.9905–1.0092
Site 9	0.9935–1.0095	0.9935–1.0046	-	0.9936–1.0083
Site 4	0.9942–1.0084	1.0002–1.0086	0.9936–1.0081	0.9946–1.0081
Site 6	0.9924–1.005	0.9895–1.0118	-	0.9937–1.0033
Site 7	0.9964–1.0039	0.9844–1.021	-	0.9914–1.0094

present a comparison of the RRIs obtained by the present study, along with the corresponding 95% CI (confidence interval), with similar studies conducted around the world. Table 2 shows that the 95% CI values for the RRI obtained for SO₂ by the present study in Mexico City were very similar to those reported for other locations around the world. The RRI values obtained by the present study for people >60 years in Mexico City were lower than those found by other studies in Barcelona and Zaragoza, Spain, and Chongqing, China. When respiratory and circulatory causes of death were considered, the same trend was identified, wherein the RRI values obtained in Mexico City were lower than those reported for Barcelona and Zaragoza, Spain, Lyon, France, and Sao Paulo, Brazil.

Table 3 shows that the 95% CI RRI values obtained for CO in Mexico City by the present study were very similar to those reported for other locations around the world. The values obtained for people >60 years were similar to those for the city of Monterrey, Mexico, but were lower than those found for cities in Spain, Europe, and China. The same trend was found when respiratory and circulatory causes of death were considered, with values obtained that were very similar to those found for Monterrey, Mexico, but lower than those reported for cities in Spain, Italy, and China, and the city of Sao Paulo, Brazil.

Table 4 shows that the 95% CI RRI values obtained for NO₂ by the present study in Mexico City for people >60 years were lower than those found in cities in both Canada and China and were much lower than those found in Panama City. For respiratory causes of death, the 95% CI RRI values found in the municipalities of Mexico City were lower than those reported for Canadian and Italian cities and Panama City, while they were similar to those reported for Monterrey, Mexico, and cities in China. The values obtained by the present study for circulatory causes of death in Mexico City were similar to those found in Monterrey and cities in China and lower than those reported for cities in Italy and Canada, Panama City, and North Korea. Finally, the RRIs obtained by the present study for all causes of death in the municipalities of Mexico City were, in all cases, lower than those reported for London, Tehran, Denmark, and cities in China and France, as well as those reported by the DUELS project.

Table 5 shows that the 95% CI RRI values obtained for O₃ by the present study in Mexico City for people aged over 60 years were lower than those found in cities in Canada as well as in Stockholm and Panama. The RRI values obtained for respiratory causes of death by the present study were similar to those found in Monterrey, lower than those reported in cities in Canada and Italy, and Panama City. The 95% CI values for circulatory causes of death found in Mexico City by the present study were similar to those found in Monterrey, lower than those reported in Panama City and cities in Canada, Italy, and China. Finally, the RRI values found by the present study for all causes of death were similar to those reported for some cities in Europe, China, and Canada and lower than those reported for some cities in England, as well as South London and Hefei, China.

Table 6 shows that the 95% CI RRI values obtained for PM₁₀ by the present study in Mexico City in the case of people older than 60 years were similar to those found in North Korea, Stockholm, Tehran, and cities in Canada, but lower than those reported for Panama City. The RRI values obtained for respiratory causes of death in the municipalities of Mexico City were similar to those reported for cities in Italy, but lower than those found in Panama City. The RRI values found for circulatory causes of death by the present study were similar to those found in Monterrey, Pudahuel and Temuco in Chile, Kermonshah in Iran, and cities in Italy, but lower than those found in Panama City. Finally, the RRI values found by the present study for all causes of death in Mexico City were similar to those found in Kermonshah, Iran.

4. Conclusions

The SO₂ concentrations obtained by the present study were uniform, except for sites 7, 8, 9, and 11, with site 11 (Azcapotzalco) corresponding to the municipality presenting the highest concentrations of this pollutant. The CO concentrations obtained were highly variable, with sites 4, 7, 9, and 11 corresponding to the municipalities presenting the highest levels. A high level of variability was observed for NO₂ concentrations, an expected finding as these depend on the peak hours of vehicular traffic, with the highest concentrations found at sites 7, 8, 9, 10, and 11. The variability observed in O₃ levels was lower than that observed for CO and NO₂, with the highest levels found at sites 1, 3, 5, and 6. Finally, the highest PM₁₀ levels were observed at sites 4, 7, 8, and 11.

Except for ozone, all criteria air pollutants tested presented their highest levels at sites 7, 9, and 11, which correspond to Venustiano Carranza, in the north of the city, Iztacalco, in the center of the city, and Azcapotzalco, located in the north of the city. Therefore, it can be concluded that the areas of Mexico City in which air quality can be considered unacceptable in terms of the official standards are located in the center and north. The concentrations of O₃ presented a different spatial distribution, with the highest levels found in the municipalities of Alvaro Obregon (located in the center area), Cuajimalpa (located in the southwest area), Xochimilco (located in the south of the city), and Tlahuac (located in the southeast of the city). Therefore, it can be concluded that the areas of Mexico City in which the air quality was unacceptable in terms of ozone concentrations were located in the center and the south.

Mortality during the study period presented a seasonal pattern, with the highest number of deaths occurring during the cold months and the winter season. While site 11 (Azcapotzalco) and site 7 (Venustiano Carranza) presented the highest mortality rates, site 4 (Iztapalapa) and site 10 (Gustavo A. Madero) contributed in great proportion to total mortality in Mexico City. By age group, the daily mortality values were higher in the population aged over 60 years, suggesting that this group was the most vulnerable. Women presented a higher number of deaths than men, while, by specific cause of death, women presented a higher level of mortality by circulatory causes of death than men, who presented a higher number of deaths due to respiratory causes.

Most of the sites studied exceeded the maximum permissible limit established by the Mexican standards [33] for SO₂, with site 11 (Azcapotzalco) presenting the highest number of values above the limit. All sites exceeded the values established by the international air quality standards (WHO and U.K. regulations). No municipality exceeded the maximum permissible limit established by the Mexican regulations [30] for CO, although the CO limit established by the WHO was exceeded in all municipalities. Except for sites 1 and 6, all municipalities exceeded both the established limit for NO₂ by the Mexican [31] and the international standards; an expected finding as the values established by the international standards are very close to those established in the Mexican regulations [31]. No municipality exceeded the maximum permissible limit established in Mexico for ozone [29], which coincides with the U.S. EPA air quality standard value. Finally, except for site 3, all sites studied exceeded the maximum permissible limit for PM₁₀ [32] with values that also exceeded the limits established in both the U.K. regulations and the WHO standards.

The PCA and RLM results showed that sites 4, 7, 9, and 11 presented the highest R² values. The estimation of the RRI values revealed that site 11 (Azcapotzalco) presented a high-level association between daily mortality and NO₂, O₃, and PM₁₀ concentrations for people aged over 60 years. Site 7 (Venustiano Carranza) presented a significant association between daily mortality and SO₂, CO, O₃, and PM₁₀ concentrations for respiratory diseases. The NO₂ concentrations showed a significant association with mortality, by all causes of death, at site 9 (Iztacalco) for people aged over 60 years.

In terms of the hypothetical 10% increase in SO₂ levels, an increase of 0.22% was projected for site 10 (Gustavo A. Madero) and an increase of 0.18% for site 11 (Azcapotzalco). The highest projected increases in daily mortality for this hypothetical scenario for CO concentrations were at site 7 (Venustiano Carranza), site 8 (Cuauhtemoc), site 10 (Gustavo A. Madero), site 11 (Azcapotzalco), site 9 (Iztacalco), and site 4 (Iztapalapa). Although no municipality was projected to exceed the maximum permissible limit established in Mexico for CO [30], the projections for all municipalities exceeded the international air quality standards, suggesting that the Mexican regulations for CO should be reviewed, as all sites also presented significant RRI values. The hypothetical 10% increase in NO₂ levels resulted in a significant increase in daily mortality at site 9 (Iztacalco), site 2 (Coyoacan), site 10 (Gustavo A. Madero), and site 4 (Iztapalapa). However, the hypothetical scenario for ozone levels did not correspond to a significantly increased risk in mortality in Mexico City. This finding coincides with the air quality results obtained in the present study, as no municipality exceeded the maximum permissible limit established for ozone [29]. Finally, a hypothetical 10% increase in the daily mean concentrations of PM₁₀ presented a significant projected increase in the risk of mortality for site 11 (Azcapotzalco), site 4 (Iztapalapa), site 1 (Alvaro Obregon), site 7 (Venustiano Carranza), and site 3 (Cuajimalpa).

It should be noted that the RRI values for PM₁₀ were the highest found for the hypothetical scenario considered by the present study, suggesting that the Mexican government should take action to reduce this risk. In this regard, during the 2019–2021 period, all regulations in Mexico were updated with stricter maximum permissible limits, although these levels remain below those recommended by the WHO. Therefore, the present study should be repeated in Mexico City, in order to take into account the limits established by the updated regulations. Finally, decision makers will be able to use the data reported in the present study to propose or improve the current regulations, programs, or actions focused on protecting the population aged >60 years against the effects of atmospheric pollution in Mexico City.