📄 medicine

Urban Mobility and Population Health in The Hague: A Multi-Scale Analysis

This study of The Hague reveals that while aggregated regional mobility patterns show a modest inverse association with population disease burden after adjusting for contextual factors, this relationship weakens significantly at the neighborhood level, suggesting that mobility data is better interpreted as an indicator of urban connectivity and population composition rather than a standalone measure of health.

Original authors: Hielke Muizelaar, Marcel Haas, Rimke Vos, Ilonca Vaartjes, Ester de Jonge, Lampros Stergioulas, Jessica Kiefte-de Jong, Marco Spruit

Published 2026-10-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Hielke Muizelaar, Marcel Haas, Rimke Vos, Ilonca Vaartjes, Ester de Jonge, Lampros Stergioulas, Jessica Kiefte-de Jong, Marco Spruit

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Cities are living organisms, constantly pulsing with the movement of people traveling to work, to shops, to friends, and to doctors. For decades, public health experts have understood that where you live shapes your health, but a newer question has emerged: can the way people move through a city tell us something about their well-being? The idea is that mobility—the distance and frequency with which people travel—might act as a mirror, reflecting how easily residents can access the resources they need to stay healthy. If a neighborhood is full of people traveling far and wide, does that mean they are healthier, or perhaps that they have better access to transport? Or is it simply that the people living there are different in ways that have nothing to do with the movement itself? This question sits at the intersection of geography and medicine, exploring whether the simple act of moving through an urban landscape carries a hidden message about the health of the population.

In a recent study focused on The Hague, a major coastal city in the Netherlands, researchers set out to test this idea using a massive dataset of mobile phone movements. They wanted to know if the average distance people traveled from their homes could serve as a reliable indicator of population health, specifically looking at chronic conditions like heart disease, diabetes, and the use of multiple medications. The team, led by scientists from Leiden University Medical Center and other institutions, analyzed data from January to July 2019, a period before the global pandemic changed how people moved. They looked at eight distinct zones within the city, measuring how far people traveled on average each day. They then compared these movement patterns against detailed health records and social data from forty-four smaller neighborhoods within those zones. The goal was to see if the distance people traveled was linked to lower rates of disease, or if that link disappeared once other factors like income, housing, and education were taken into account.

The researchers found that movement patterns in The Hague were indeed very different from one zone to another. In some areas, the average outgoing trip was about 9.5 kilometers, while in others, it stretched to 12.7 kilometers. At first glance, a pattern seemed to emerge: zones where people traveled farther tended to have lower rates of chronic disease. For instance, areas with longer average travel distances showed fewer people using medication for high blood pressure, cholesterol, or diabetes. This initial observation suggested a hopeful connection: perhaps the ability to travel further is a sign of a healthier, more connected community. However, the story became much more complex when the researchers looked closer at the people living in those areas.

When the team adjusted their analysis to account for the social and environmental conditions of each neighborhood, the direct link between travel distance and health began to fade. They realized that the people who traveled further were not necessarily healthier because they moved more; rather, they were healthier because they lived in neighborhoods with specific characteristics. These characteristics included higher incomes, better access to green spaces, and a population composition that included a higher proportion of residents with Western migration backgrounds. In fact, the study found that the background of the people living in a zone was a stronger predictor of how far they traveled than almost any other factor. Once the researchers mathematically removed the influence of these social and demographic factors, the remaining "extra" movement did not seem to have any independent connection to health. In other words, the movement itself was not the cause of the health difference; it was simply a side effect of the type of neighborhood and the people living there.

The study also highlighted a crucial lesson about how we measure things in cities. The researchers compared data collected at two different scales: the eight large zones used for tracking phone movements, and the forty-four smaller administrative neighborhoods used for health records. When they looked at the big zones, the link between travel and health appeared moderate. But when they tried to apply that same link to the smaller neighborhoods, the connection vanished. This suggests that using a broad measure of movement to predict health in a specific small area can be misleading. The data showed that aggregated movement is better understood as a map of the city's social structure and connectivity rather than a standalone health metric. It tells us about the composition of the population and how the city is organized, but it does not tell us directly if a specific group of people is healthy or sick.

Ultimately, the findings suggest that while mobile phone data offers a fascinating glimpse into how people interact with their cities, it cannot replace traditional health indicators. The distance people travel is deeply embedded in the social and economic fabric of their neighborhoods. A long commute might look like a sign of a healthy, active life, but in this study, it was more likely a reflection of who lives there and what resources they have access to. The researchers concluded that to truly understand urban health, we must look at the whole picture: the movement, the people, and the place they call home. Mobility is a powerful tool for understanding how a city functions, but it is not a magic key that unlocks the secrets of population health on its own.

Technical Summary: Urban Mobility and Population Health in The Hague

Problem Statement
Urban mobility data, particularly aggregated mobile-phone-derived metrics, are increasingly utilized to infer population health and resilience. However, mobility patterns are deeply embedded within broader socioeconomic, environmental, and demographic contexts. A critical methodological challenge arises when mobility and health indicators are measured at different spatial scales (e.g., functional mobility regions vs. administrative neighborhoods). It remains unclear whether aggregated mobility provides independent, health-relevant variation beyond the shared contextual conditions that shape both movement and disease burden. Furthermore, the interpretation of mobility-health associations is complicated by the "modifiable areal unit problem," where the choice of spatial units influences observed relationships. This study addresses whether aggregated urban mobility retains an association with population health after accounting for co-occurring contextual factors, specifically examining the impact of spatial scale and population composition.

Methodology
The study employs an ecological, cross-sectional, multi-scale analysis conducted in The Hague, the Netherlands, covering January to July 2019.

Data Sources:
- Mobility: Aggregated daily origin-destination flows from the Vodafone Mezuro dataset, covering eight provider-defined "Mezuro regions." The primary metric was the mean outgoing distance (weighted by movement volume) per region. A secondary metric, outgoing mobility rate, was used for sensitivity.
- Health Outcomes: Data from the ELAN (Extramural Leiden University Medical Center Academic Network) primary-care infrastructure, aggregated to the neighborhood level. Indicators included cardiometabolic medication use (lipid-lowering, glucose-lowering, blood-pressure-lowering), polypharmacy (≥5 and ≥10 unique ATC-4 codes), and a syndemic-based disease burden measure (predefined interacting condition combinations).
- Contextual Covariates: Variables from Statistics Netherlands (CBS), the RIVM Health Monitor, and the Klimaateffectatlas, grouped into conceptual domains: age composition, socioeconomic position, education, migration background, housing, lifestyle/social factors, facilities access, physical environment, and social vulnerability.
Spatial Alignment:
- Mobility data existed at the coarse scale of eight Mezuro regions.
- Health and contextual data existed at the finer scale of 44 administrative neighborhoods.
- Neighborhoods were assigned to the Mezuro region with the greatest geographic overlap (median 98.6% overlap) to create a cross-scale dataset.
Analytical Strategy:
- Residual-Based Adjustment: The core analytical framework involved modeling mobility and health outcomes separately as functions of the same selected contextual domains.
  1. Mobility Model: Mean outgoing distance was regressed against contextual variables to generate "residual mobility" (mobility variation independent of context).
  2. Health Model: Each health outcome was regressed against the same contextual variables to generate "residual health" (health variation independent of context).
- Association Analysis: Spearman's rank correlation ( $\rho$ $ρ$ ) was calculated between residual mobility and residual health.
  - Regional Level: Correlations were computed across the eight Mezuro regions.
  - Cross-Scale Neighborhood Level: Correlations were computed between the regional mobility residuals and neighborhood health residuals.
- Uncertainty Estimation: Due to the limited number of independent mobility observations (8 regions), 95% confidence intervals were derived using 1,000 cluster-bootstrap resamples, treating entire Mezuro regions (and their constituent neighborhoods) as the resampling unit. Leave-one-region-out (LORO) analyses were performed to assess sensitivity to individual regions.

Key Results

Contextual Drivers of Mobility: Outgoing mobility varied significantly across the eight regions and was strongly patterned by contextual factors. Lifestyle indicators (specifically adherence to physical activity guidelines) showed the strongest explanatory power ( $R^2 = 0.742$ ), followed by education ( $R^2 = 0.493$ ). Notably, Western migration background was strongly associated with outgoing mobility distance at the regional level ( $R^2 = 0.893$ ), though this was treated as an area-level compositional association.
Regional-Level Associations: At the Mezuro-region level, higher-than-expected mobility (after contextual adjustment) tended to coincide with lower-than-expected disease burden. Correlations were moderate and negative, ranging from $\rho = -0.26$ (diabetes medication) to $\rho = -0.48$ (syndemic measure).
Cross-Scale Neighborhood Associations: In the cross-scale analysis, residual correlations between mobility and health outcomes were considerably weaker and statistically indistinguishable from zero.
- Correlations ranged from $\rho = -0.11$ (cholesterol medication) to $\rho = 0.10$ (syndemic measure).
- Cluster-bootstrap 95% confidence intervals included zero for all outcomes (e.g., Syndemic measure: $[-0.18, 0.40]$ ).
- LORO analyses confirmed that no single region drove these weak associations.
Sensitivity Analyses: Results were robust to alternative mobility measures (outgoing rate), stricter polypharmacy definitions (≥10 meds), and substitution of contextual variables within domains. The inclusion of Western migration background as a specific adjustment variable altered model fit but did not yield strong residual mobility-health associations.

Key Contributions

Multi-Scale Framework: The study explicitly addresses the challenge of linking data across different spatial supports (functional mobility regions vs. administrative neighborhoods), demonstrating how geographic units influence the interpretation of mobility-health relationships.
Residual-Based Approach: By modeling mobility and health as functions of shared contextual domains, the study isolates whether mobility retains independent health-relevant information. The findings suggest that aggregated mobility largely reflects the same underlying social and environmental structures that drive health outcomes, rather than acting as an independent predictor.
Syndemic Perspective: The research incorporates a syndemic-based disease burden measure alongside traditional polypharmacy and medication-specific indicators, offering a multidimensional view of co-occurring conditions in relation to mobility.
Population Composition Insight: The analysis highlights that aggregated mobility indicators are hybrid measures reflecting regional connectivity, environmental context, and population composition (specifically migration background), cautioning against interpreting them solely as individual behavioral metrics.

Significance and Claims
The paper concludes that aggregated urban mobility should not be interpreted as a standalone indicator of population health. Instead, it functions as a "spatially and contextually embedded indicator of urban connectivity and population composition."

The study claims that the contrast between moderate regional-level associations and negligible cross-scale neighborhood associations underscores the importance of spatial scale. While mobility data can reveal broad patterns of urban connectivity, they do not appear to capture additional health-relevant variation beyond the measured contextual conditions (socioeconomic, environmental, and demographic) at the neighborhood level. Consequently, the authors argue that differences in spatial resolution should be treated as part of the substantive interpretation of mobility data, rather than merely a technical limitation. The findings suggest that for chronic disease burden, aggregated mobility is best understood as a reflection of the broader urban context and population structure, with limited utility as an independent predictor of neighborhood-level disease burden after accounting for these factors.

Technical Summary: Urban Mobility and Population Health in The Hague

More like this