Cardiovascular health in perspective: a comprehensive five-year geodatabase of hospitalizations and environmental factors in Mashhad, Iran

Abstract

Objectives

This data note presents a comprehensive geodatabase of cardiovascular disease (CVD) hospitalizations in Mashhad, Iran, alongside key environmental factors such as air pollutants, built environment indicators, green spaces, and urban density. Using a spatiotemporal dataset of over 52,000 hospitalized CVD patients collected over five years, the study supports approaches like advanced spatiotemporal modeling, artificial intelligence, and machine learning to predict high-risk CVD areas and guide public health interventions.

Data description

This dataset includes detailed epidemiologic and geospatial information on CVD hospitalizations in Mashhad, Iran, from January 1, 2016, to December 31, 2020. It contains 52,176 confirmed CVD cases and includes demographic information such as age, gender, admission date, ICD-10 codes, occurrence of death, and length of hospital stay. The median age of patients was 64 years, with 54.44% male. A notable 9.41% of patients died during hospitalization. In addition to the CVD hospitalization case file and its shape file created by joining with 1301 census tracts, this dataset includes environmental factors such as air quality indicators (SO₂, PM₂.₅, CO, etc.). It also incorporates socio-economic variables (population density, illiteracy, and unemployment rates), public infrastructure, and built environment data, providing a comprehensive view of cardiovascular health in Mashhad.

Objective

Cardiovascular disease (CVD) remains one of the leading causes of mortality worldwide, responsible for a considerable proportion of global deaths each year [1]. The growing burden of CVD highlights the need to understand not only its biological risk factors but also the broader environmental and socioeconomic determinants that contribute to its burden. Environmental factors, such as air pollution, socioeconomic conditions, and the built environment, are increasingly recognized for their role in exacerbating cardiovascular conditions [2, 3]. Urban environments pose distinct challenges, where factors such as high population density, increased exposure to pollutants, and restricted access to green spaces can amplify the risks associated with cardiovascular disease [4]. Thus, studying the spatial distribution of CVD in conjunction with these modifying factors is crucial for addressing this global health challenge.

This data note presents data on the geographic distribution of hospitalized CVD patients in Mashhad, Iran. It also outlines potential environmental variables, such as air quality, green spaces, and urban density, that may be associated with cardiovascular health outcomes. To achieve this, we systematically collected, integrated, and compiled a comprehensive geospatial and epidemiologic dataset of hospitalized CVD patients across a five-year period. By leveraging this extensive dataset, which offers high spatial and temporal granularity, researchers can conduct various analyses to uncover spatial patterns of CVD and explore the relationship between environmental stressors and cardiovascular risk [5, 6].

The dataset offers significant potential for research applications, including the identification and prediction of high-risk CVD areas through spatial epidemiology, artificial intelligence, and machine learning. Moreover, this geodatabase can significantly inform urban planning and environmental policy by providing insights into how urban design, land use, and transportation systems influence cardiovascular health. By identifying areas with high concentrations of pollutants or limited access to healthcare services, urban planners can develop targeted interventions to mitigate health risks. Additionally, the data can support the formulation of evidence-based policies aimed at improving public health outcomes, such as enhancing green spaces, optimizing public transportation routes, and implementing stricter environmental regulations.

Data description

Table 1 shows the details of the four data files plus the data dictionary. The dataset encompasses detailed geospatial and epidemiologic information on CVD hospitalizations in Mashhad, Iran, collected from January 1, 2016, to December 31, 2020. The data were sourced from the Hospital Information Systems (HIS) managed by Mashhad hospitals, which serves as a comprehensive repository of information for patients admitted to hospitals. The dataset includes hospitalizations for various cardiovascular disorders, classified using ICD-10 codes, including ischemic heart disease (I20-I25), atrial fibrillation (I48), heart failure (I50), ischemic stroke (I63, I65-I67, G45, G46), and atherosclerotic disease (I70-I77.1).

A series of preprocessing steps were implemented to refine the initial raw dataset. Address fields were manually corrected for spelling and completeness. Records were excluded based on specific criteria: patients residing outside Mashhad, those staying at hotels or inns due to the city’s prominence as a pilgrimage site, and any records with missing address information. These exclusion criteria ensure that the dataset accurately reflects the local population and enhances the robustness of subsequent analyses. After preprocessing, the final dataset comprises 52,176 hospitalization records, including details such as age, sex, admission dates, ICD-10 codes, length of stay (LOS), and in-hospital mortality status (Data file 1). The dataset comprises 54.44% male patients, with a median age of 64 years (interquartile range (IQR): 53–76 years) across both sexes. The median LOS in the hospital was 2 days (IQR: 1–5 days). A notable 9.41% of the CVD patients succumbed to the disease, resulting in 4,907 deaths during the study period. While 2019 recorded the highest level of CVD admissions, Joinpoint Regression analysis revealed a significant shift in trends starting in late 2019, coinciding with the onset of COVID-19. From November 2019 to December 2020, CVD rates declined notably, with monthly percent changes (MPC) of -6.04 for males and − 6.49 for females, likely due to reduced healthcare access and changes in health behaviors during the pandemic [7, 8].

To create a geospatial dataset, patient addresses were geocoded and mapped using the Google My Map service. To enhance confidentiality, a 100-meter jittering technique was employed, randomly displacing each geocoded point within a 100-meter radius to obscure exact locations while preserving spatial accuracy [9]. All geocoded cases were then spatially joined with the census tracts (n: 1301) for detailed geographic analysis. After excluding 44 cases outside the geographical boundaries, the final CVD geodatabase comprised 52,132 cases (Data file 2). This rigorous approach to geocoding enhances the dataset’s robustness, ensuring reliable spatial analysis. Please refer to Supplementary Material 1 and 2 for heat plots on CVD hospitalizations, deaths, LOS, and the spatial distribution of Standardized Incidence Ratios (SIR) for CVD hospitalizations during the study period.

The study area spanned 1,301 census tracts, covering a total population of 3,001,225 individuals, based on the 2016 population and housing census (male: 50.11%). In addition to the CVD hospitalization data, this dataset also includes environmental factors relevant to cardiovascular health, which are provided as a geographical shape file (Data file 3). Rationale for selecting these environmental variables was grounded in previous literature that has examined the association between CVD and environmental factors [10,11,12,13,14,15]. Access to these variables from various sources was crucial for our analysis. These factors encompass air quality indicators, such as sulfur dioxide (SO₂), particulate matter (PM₂.₅ and PM₁₀), carbon monoxide (CO), nitrogen dioxide (NO₂), and ozone (O₃) levels. The average concentration of air pollutants was measured using data collected from 22 monitoring stations across Mashhad. Census-level exposure was estimated using inverse distance weighting (IDW) interpolation, where pollutant values were calculated for grid rectangles based on the proximity to nearby monitoring stations [16]. Finally, overall pollutant averages for each census were derived from the interpolated grid values within each census tract. Public infrastructure data, including the distribution of transportation facilities such as bus stops, fuel stations, and road intersections, is included. Additionally, built environment data, incorporating the Normalized Difference Vegetation Index (NDVI), provides information on green spaces and densities of commercial and industrial centers. All data is sourced from the municipal council. Socio-economic variables, such as population density, illiteracy rates, and unemployment rates—derived from the 2016 population and housing census—are also incorporated to offer insights into the broader environmental and socio-economic context of cardiovascular disease. Notably, data files 2 and 3 can be joined using the “Tractid” feature. Data file 4 includes the data usage agreement that researchers must accept to access Data file 1. Table 1 presents a summary of the dataset, which has been made available on the Harvard Dataverse [17].

Table 1 Overview of data files

Limitations

• The use of hospital admission data inherently limits the scope of our dataset, as it may lead to an underestimation of the true incidence of CVDs. This limitation arises from the exclusion of patients treated in outpatient settings or those who did not seek medical care. Consequently, the results may not fully capture the impact of CVDs in the broader population, highlighting the need for caution when interpreting the results.

• Although we analyzed CVD trends over a five-year period, the study period overlaps with the COVID-19 pandemic, which likely influenced temporal patterns in CVD hospitalizations and outcomes. This overlap may affect the generalizability of the results to non-pandemic periods.

• Future research should expand to include data from outpatient and primary care facilities and assess the long-term effects of the COVID-19 pandemic on CVD trends in the region.