Covid-19 y análisis espacial en México

Humberto Charles-Leija


Esta investigación propone utilizar análisis exploratorio de datos espaciales como herramienta complementaria en la definición de territorios prioritarios para atención gubernamental durante la contingencia originada por el covid-19. El análisis se realiza para todos los municipios de México utilizando datos del censo 2020 y datos abiertos de epidemiología, hasta febrero de 2021. El estudio identifica agrupaciones de municipios que muestran una mayor incidencia de contagios desde que inició la pandemia, así como en la última semana. Los espacios de mayor interés se ubican en el centro del país y es ahí donde se propone realizar los mayores esfuerzos de contención.


Introduction The COVID-19 pandemic has caused the death of millions of people in the world. Mexico is one of the countries with the most deaths. The impact has been not only health but also economic and social. The Mexican government's strategy has been based on one epidemiological traffic light per state. However, some states in the country are very large territories and have municipalities that are not close to each other. On the other hand, there are regions that belong to different states and do have proximity. In this way, it may be convenient to complement the health strategy taking into account municipal data, and not only state data. The present study proposes to use Exploratory Spatial Data Analysis (ESDA) as a complementary tool in the definition of priority territories for attention to public policy in Mexico, regarding covid-19 crisis. The ESDA offers the possibility of initially identifying hot spots, that is, points where the variables of interest have the highest values and where an action by an external agent, in this case a health program, can have the greatest impact. Methodology The analysis is carried out for the 2,469 municipalities in Mexico. The geographic contiguity data were obtained from the 2020 Population and Housing Census, using shp files that the National Institute of Geography and Statistics (INEGI in Spanish) has available for the population on its website. The data referring to daily infections are from the Official page of the General Directorate of Epidemiology in Mexico. Statistical language, R, is used for data processing and Geoda for exploratory spatial analysis of data and visualization. Although the census data allow a greater level of disaggregation, epidemiological sources only allow reaching the municipal level. The central hypothesis of the study is the following: there is a spatial correlation of infections. To make spatial estimates, it is necessary to previously establish a matrix of spatial weights (w). This matrix describes whether there is a neighborly relationship of one region with the rest. Based on this, it is evaluated whether there is a contagion effect in that neighborhood. For the current case, the matrix that is constructed uses the type "queen" of the first order, that is, it considers as neighbors those municipalities that have a line or vertex of adjoining the observation, the advantage of using this criterion is that it considers as neighbors municipalities that have both vertex and extended border (LeSage, 1998). Other possible criteria are "rook" type where contiguity is defined by sharing a border and distance, determined by a maximum value of distance between observations. It is considered that queen-type contiguity can be convenient because it generates greater connectivity between nearby municipalities. In this research, an ESDA is carried out with the purpose of describing and visualizing spatial distributions, to identify atypical locations, discovering associations (spatial autocorrelation) and structures in geographical space (spatial heterogeneity) (Chasco Yrigoyen, 2003). Results Mexico has almost 2,500 municipalities. Among the 25 most affected municipalities, which correspond to about 1% of the total are the 16 that make up Mexico City and municipalities with a small population in Nuevo León, Coahuila, Chihuahua, Sonora, Baja California Sur and Yucatan. Results shows that the rate of infections does not follow a random distribution in the municipalities of Mexico. To test the hypothesis of spatial independence of a variable, the Moran Index (I-Moran) is used, this is a scalar value that can reach values between -1 and 1. When it approaches -1 it indicates that there is a negative spatial correlation, when it approaches 1 it represents a positive spatial correlation and when approaching zero it indicates spatial independence of the variable in question (Rupasingha and Goetz, 2007). The values of univariate global Moran I show a strong positive spatial autocorrelation, both for the entire period and for the last week of data analyzed. A first-order queen-type contiguity matrix was used for this analysis. When evaluating the I-Moran with data on the rate of infections in the municipalities of Mexico, a value of 0.515 was obtained, with a significance level of 99% (p-value=0.0001), from 9999 permutations. The above supports the hypothesis of the present work, there is a spatial correlation between the rates of contagion of the municipalities of Mexico. The study identifies clusters of municipalities that show a higher incidence of infections since the pandemic began, as well as in the last week. The groupings of municipalities in the last week should be the priority areas for government intervention. The results show that ESDA can be used to identify priority areas of attention to medical, social and economic needs. The analysis can be replicated in other countries at more disaggregated levels for greater targeting of efforts. The results show the proposal of priority areas for government attention based on the data available between February 5 and 12, 2021. According to this analysis, it is proposed to prioritize efforts in Mexico City and some municipalities of the State of Mexico since they are the areas with the highest number of recent infections, as well as areas that are highly affected in economic matters. Other candidate municipalities for priority are located in the states of Guanajuato, Nuevo León, Coahuila and Baja California Sur. Conclusions It is possible to generate significant impacts in the reduction of contagion rates from properly targeting the available resources. In Mexico there is a daily monitoring of the conditions of the pandemic caused by covid-19, however, the health strategy can still be improved to give priority in the regions where the crisis is most severe. The present study confirmed the hypothesis raised, there is a spatial correlation in the rate of infections in Mexico. Based on this, it is proposed that spatial data analysis can contribute to establishing a health strategy based on spatial aspects. Among the findings of the present study are that, taking into account the data until February 2021, when Mexico was just emerging from the most severe wave of infections it has experienced during the health crisis, municipalities in Mexico City and the State of Mexico were the territories of greatest interest. The above, considering the high rates of contagion in these municipalities, on the one hand, but also, integrating into the analysis the rates of contagion of the neighboring municipalities. The research was proposed from the perspective of regional economics, using a spatial analysis tool. A deeper exploration of epidemiological factors is needed to optimize the criteria for action. However, the present work may be useful for future research in epidemiology to draw on tools used by regional economists. Likewise, future studies can classify populations according to their willingness to receive the vaccine (Becchetti and Salustri, 2021) or focus on identifying the effect of sociodemographic variables on the rate of infections, deaths, etc., in this way the impact of government actions can be more clearly distinguished. Data on population density, overcrowding, etc. can be used. As well as focusing the analysis on particular regions of interest.

© Revista de estudios regionales 2014 Universidades Públicas de Andalucía