Abstract
Systems for monitoring population-level diet and nutritional intake have been considered insufficient across many countries. Recently, internet search query data have been used to examine spatial and temporal patterns of public behavior to inform public-health campaigns, policies, and interventions. Seasonal trends in public interest in behavioral change associated with obesity have been documented using such data. However, it has not been validated whether search query data can be related to diet and nutritional intake at the population level. The purpose of this study was to investigate whether trends in search query data related to behavioral changes associated with obesity reflects population nutritional intake and dieting behavior. First, long-term (2004 to 2016) trends in Australian Google search behavior were examined for the terms “weight loss”, “diet”, and “fitness” to establish monthly patterns in relative search volume (RSV). Second, monthly total energy (kJ), macronutrient, and food intake of the Australian population, and the percentage of self-reported dieters, were quantified using data collected as part of a 2011–2012 national-level survey. The two independent data sets were then compared to ascertain similarities in trends. There were distinct patterns in RSV across months, which was significantly higher than the mean during January, and lower during December, for all search terms. The decline in RSV was not linear, however, as there were significantly lower RSVs for terms during May to July, and significantly higher from August to October. Likewise, nutritional data showed a seasonal pattern, with the energy intake of survey participants highest in December and lowest in February, and the percentage of self-reported dieters closely followed monthly patterns in RSV. The proportion of energy from protein was consistent across months examined; however, energy from lipid and carbohydrate + fiber, was variable between months. Likewise, consumption patterns of different food groups was variable across months. Our analysis suggests that search query data can be used to surveil and predict dietary behavior at the population level, which has implications for producing novel and contemporaneous health information and marketing strategies.
Highlights
There is strong evidence of a global obesity pandemic related to over-nutrition and underactivity (James et al, 2001; Swinburn et al, 2011; Stevens et al, 2012)
We found that energy intake from June 2011 to May 2012 varied between months, and that the trends in energy intake closely matched the pattern in Google Trends relative search volume (RSV), especially from October through May (Table 3; Fig. 5)
We examined the digital mapping of online behavior with dietary habits to show that monthly patterns in public interest in behavior change as measured using search query data can be associated with population-level patterns in energy, macronutrient, and food intake
Summary
There is strong evidence of a global obesity pandemic related to over-nutrition and underactivity (James et al, 2001; Swinburn et al, 2011; Stevens et al, 2012). Internet search query data is one form of Big Data that has been used to provide insight into temporal and biogeographic patterns of public behavior (Brownstein et al, 2009; Nuti et al, 2014). Such data have been used to aid researchers, practitioners, and policy makers in designing spatially and temporally specific public-health campaigns, policies, and behavioral interventions (Carr and Dunsiger, 2012)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have