Infodemiology of metabolic syndrome
Analysis of population web search behaviour for complementary epidemiological surveillance
Complementary population surveillance through web search behaviour
Metabolic syndrome is advancing worldwide, but conventional surveillance detects its trends late and does not capture population interest or its geographical differences.
Metabolic syndrome is a clinical entity characterised by central obesity, insulin resistance, hypertension and dyslipidaemia, and it is a growing global public health problem. Conventional epidemiological surveillance systems detect its effects but do not capture how the population perceives risk, seeks information or understands its own health.
Complementary methods were needed to monitor population interest in metabolic syndrome and in five search topics linked to its components—hypertension, diabetes, central obesity, cholesterol and triglycerides—to inform health education strategies. The challenge was to analyse 17 years of global search data through Google Trends, assess temporal and geographical patterns, and interpret the results in light of the limitations inherent to digital epidemiology, such as the digital divide, content filtering and socioeconomic differences in internet access.
17 years
of web search data analysed at global scale (2004-2020)
6 search topics
studied: metabolic syndrome and five of its components
3 levels
of development compared across countries, following the United Nations classification
Infodemiological analysis across time periods and geographies
Digital epidemiology with global longitudinal data
Systematic extraction of normalised relative search volume (RSV) from Google Trends on a 0-100 scale. Time series analysis over three periods to assess trends using regression and Pearson correlation, together with stationarity testing of the series with the KPSS test.
Statistical analysis with biases put in context
Non-parametric tests (Kolmogorov-Smirnov, Kruskal-Wallis), given the non-normal distribution of the search data. Correlation analysis between topics to identify associations in population behaviour. Geographical stratification by countries' level of development, with explicit consideration of the digital divide and unequal internet access during interpretation.
Geographical and temporal visualisations
Global heat maps to identify hotspots of health interest visually. Time series by region to detect differences in how population interest has evolved. Comparative analyses of metabolic syndrome and its components to identify differences in population interest.
Metabolic syndrome is searched for far less than its components
Differences in interest between the syndrome and its components
The median RSV for "metabolic syndrome" (2/100) was significantly lower than for its components (hypertension 46/100, cholesterol 41/100, diabetes 14/100). The negative correlation with central obesity (R = -0.70; p < 0.001), together with the lower search volume for the technical term, suggests a different pattern of interest between metabolic syndrome and some of its components.
Declining trend and geographical disparities
A significant decline in interest in metabolic syndrome over the study period (Pearson coefficient = -0.71, p < 0.001), in contrast with rising trends for its components. Developed countries showed a higher search volume than developing countries (p < 0.001), a pattern consistent with disparities in health education and access to information.
A reproducible framework for complementary surveillance
A methodology that can be transferred to monitoring other chronic, emerging or re-emerging conditions, to complement conventional surveillance systems. The capability demonstrated in big data analysis of web behaviour opens up rapidly updated routes for epidemiological surveillance to inform public health strategies.
What this case shows about the way we work
Experience in applied infodemiology
Demonstrated capability in big data analysis of web behaviour for complementary epidemiological surveillance, an emerging area in Spanish public health.
Robust methodology with biases put in context
The digital divide and level of development are taken into account when interpreting search patterns, avoiding spurious conclusions in global studies.
Long time series and advanced statistical analysis
Non-parametric tests, correlation analysis between topics and KPSS stationarity testing to draw rigorous evidence from public data.
Transferable to multiple conditions
A framework that can be applied to monitoring population interest in any chronic, emerging or re-emerging disease, complementing conventional surveillance.
Conozco a Julia y Rubén desde hace años. Siempre han destacado por su compromiso y profesionalidad. No han dejado de formarse tratando de estar siempre a la vanguardia de las novedades en el campo de la ciencia de datos, aportando buenas ideas en todos los proyectos en los que hemos colaborado.
Dr Javier Sanz-Valero
Professor, Departamento de Salud Pública e Historia de la Ciencia, Universidad Miguel Hernández de Elche
Resulting scientific output
Estudio infodemiológico aplicado a la salud ocupacional de enfermedades y trastornos nutricionales
Trends and seasonality of information searches, carried out through Google, on metabolic syndrome and occupational health: infodemiological study
Trends and Seasonality of Information Searches Carried Out through Google on Nutrition and Healthy Diet in Relation to Occupational Health: Infodemiological Study
Capabilities deployed in this project
Shall we talk?
Tell us what you are working on and we will give you a straight answer on how we can help.