Study design and sampling
Data from the present study were derived from the ‘Sustainable prevention of obesity through integrated strategies’ (SPOTLIGHT) study . In this study, a cross-sectional survey was conducted among 6037 adult participants from 60 selected neighbourhoods of five urban regions across Europe: Ghent and suburbs (Belgium), Paris and inner suburbs (France), Budapest and suburbs (Hungary), The Randstad (a conurbation including the cities of Amsterdam, Rotterdam, The Hague, and Utrecht in The Netherlands) and Greater London (UK). Neighbourhoods were defined according to small-scale local administrative boundaries. Sampling of neighbourhoods and recruitment of participants has been described in detail elsewhere . After completing the survey, Dutch and Belgian participants who indicated being interested in future studies, were contacted to participate in an accelerometer study. In total, 255 Dutch and 167 Belgian participants (participation rate = 16%) of 24 different neighbourhoods (i.e. 12 Dutch and 12 Belgian neighbourhoods; mean surface is 1.2 km2 and 1.4 km2 respectively) agreed to wear an accelerometer for seven consecutive days. In the Netherlands, the accelerometers were sent to the participants’ home addresses including a written instruction on how to wear the device. In Belgium, researchers visited the participants at home to attach the accelerometers and to provide the instructions. In total, informed consents and valid accelerometer data were obtained from 225 Dutch and 149 Belgian participants between March and August 2014. The study was approved by the VU University Medical Centre ethics committee (2012/314) and the Ghent University Hospital ethical committee (EC/2013/518).
Accelerometer-determined sedentary time
The Actigraph triaxial accelerometer (Model GT3X, Actigraph, LLC, Fort Walton beach, FL) was fixed with an elastic belt around the waist on the right side of the participants. Participants were asked to wear the accelerometer for seven consecutive days during waking hours, except during bathing and other water-based activities. A valid day was defined as a day which contained at least 10 h of accelerometer data, and only adults with at least four valid days were included in the analyses. Each minute of wear-time was classified into sedentary (<100 cpm), light (100–2019 cpm), moderate (2020–5998 cpm) and vigorous intensity activity (>5999 cpm) according to commonly used cut points for adults . Non-wear time was defined as 60 min of consecutive zeroes, allowing for two interruptions of less than 100 counts per minute .
Objectively measured physical environmental neighbourhood factors
Objectively measured physical environmental factors of the selected neighbourhoods were assessed using the reliable and valid SPOTLIGHT Virtual Audit Tool (S-VAT) . The S-VAT contains 42 items, grouped into eight domains: walking (six items), cycling (eight items), public transport (two items), aesthetics (nine items), land use mix (three items), grocery stores (five items), food outlets (six items) and recreational facility-related items (three items). All items were assessed at street segment level within the 24 selected neighbourhoods in the Netherlands and in Belgium by trained researchers of the SPOTLIGHT project team. Street segment level data were aggregated to the neighbourhood level by taking the percentage of street segments with each feature in the neighbourhood . Items for which no association with sedentary time was expected based on the SOS framework (e.g. liquor stores, type of residential buildings)  were excluded from the statistical analyses, as well as items with limited variance, i.e. if more than half of the neighbourhoods yielded the same percentage (e.g. indoor recreational facilities, take away restaurants, abandoned buildings, and railway/underground stations). As a result, 16 physical environmental neighbourhood factors were included in the statistical analyses (see Table 2).
Covariates considered in the analyses were retrieved from the online survey, and included age, gender, educational level (lower, higher), and household composition (number of people in the household).
Only participants for whom questionnaire data, S-VAT data and valid accelerometer data were available were included in the current analyses (N = 347). Descriptive statistics were computed using R software, version 3.1.2. to summarize participant characteristics. General linear mixed models were used to examine the associations between physical environmental neighbourhood factors and sedentary time, as the Kolmogorov-Smirnov test revealed a normal distribution of the dependent variable (sedentary time). A random intercept was added within all models to account for clustering of participants at neighbourhood level. Physical environmental neighbourhood items were multiplied by 100 before entering in the models to facilitate interpretation, so that a 1-unit increase represents a 1% increase in the presence of neighbourhood characteristics.
In a first step, the separate associations between the sixteen selected physical environmental neighbourhood factors and sedentary time were analysed using single regression models. In a second step (model 1), the same physical environmental neighbourhood factors were entered in a multivariable regression model, except the presence of cafés, local food shops, residential gardens, litter, bicycle lanes and public parks. As the variance inflation factors showed that these variables violated the assumption of multicollinearity, these variables were not included in the multivariable regression model. In a final step (model 2), the covariates were added to the model to control for their influence on the potential associations. Statistical significance was set at p ≤ 0.05.