Health impacts of the M74 urban motorway extension: a mixed-method natural experimental study

Study population and outcomes

To address the questions outlined in Chapter 1, Main research questions, we conducted a mixed-method longitudinal study with the elements outlined in Table 2.

We combined this with the analysis of routinely available population data on road traffic accidents and travel behaviour.

Objectives

The objectives of the study were:

1. to characterise the context, content and implementation of the intervention by means of an environmental survey

2. to follow up a cohort of residents of the intervention and control areas had who previously responded to a postal survey at baseline [2005; time point 1 (T1)]

3. to draw new repeat cross-sectional samples of residents of the intervention and control areas for a postal survey at follow-up [2013; time point 2 (T2)]

4. to objectively measure the travel and activity patterns of a subsample of survey participants

5. to estimate changes and differences in the primary and secondary outcome measures between the intervention and control areas, and by level of exposure to environmental changes within those areas

6. to examine the extent to which any changes in outcomes were mediated by changes in perceptions of the neighbourhood environment

7. to examine the extent to which any changes in behavioural outcomes were associated with changes in well-being

8. to interview a further subsample of participants to elicit how the effects of the environmental changes were experienced by local residents and how any changes in behaviour or well-being were mediated and enacted at individual and household level

9. to examine changes in the incidence and sociospatial distribution of road traffic accidents on the road network

10. to explore trends and spatial variation in travel behaviour using existing national population data sets

11. to examine the extent to which the results of the different analyses supported either of the two competing overall hypotheses regarding the cumulative effects of the intervention (see Table 1).

Study design

The study used a combination of quantitative [cohort, cross-sectional, repeat cross-sectional and interrupted time series (ITS)] and qualitative (documentary analysis and interview) research methods to evaluate both individual- and population-level changes in health and health-related behaviour, and to develop a more in-depth understanding of how these changes were experienced and brought about.

The study comprised six main components (Figure 2):

1. an environmental survey consisting of documentary analysis, interviews with key informants, field visits and use of Google Earth (Google Inc., Mountain View, CA, USA) images (objective 1)

2. core surveys of local study areas to compare changes in neighbourhood perceptions, travel behaviour, physical activity and well-being in the intervention and control areas by means of combined cohort and repeat cross-sectional follow-up surveys of local residents (objectives 2, 3, 5, 6 and 11)

3. an objective measurement study of a subsample of core survey participants to quantify any differences in physical activity between intervention and control areas (objectives 4, 5, 6 and 11)

4. a qualitative study of a subsample of core survey participants to elucidate their experiences of environmental changes and the mechanisms through which these may have influenced behaviour (objectives 7, 8 and 11)

5. an analysis of existing national population data sets to evaluate the impact of the intervention on road traffic casualties and to describe concurrent regional and national trends in travel behaviour (objectives 9–11)

6. running alongside all other components, a programme of community engagement to help shape the final study design, elicit a wider range of accounts and develop a shared understanding and interpretation of the emerging findings (objective 11).

FIGURE 2.

Study components.

Logic model

As discussed in Chapter 1, we summarised the contrasting narratives about the motorway into vignettes describing two competing overarching hypotheses about the effects of the intervention. In order to operationalise the relationships of interest and refine our analytical priorities, these were further developed into an overarching logic model describing the main putative causal relationships to be investigated at follow-up (Figure 3).

FIGURE 3.

Overarching logic model.

Study areas

At baseline, three local study areas were defined: the ‘M74 corridor’ intervention area (South) and two control areas, one of which surrounded the existing M8 and M80 motorways (East) and one of which had no comparable major road infrastructure (North) (Figure 4). These study areas were carefully and iteratively delineated at baseline using spatially referenced Census and transport infrastructure data combined with field visits to ensure similar aggregate socioeconomic characteristics and broadly similar topographical and urban morphological characteristics apart from their proximity to urban motorway infrastructure (Table 3). 55 Baseline analysis confirmed no significant differences between the achieved survey samples in these three areas on any socioeconomic or behavioural summary measures apart from a minor difference of borderline statistical significance (p = 0.053) in the distribution of housing tenure. 57 All three study areas extend from inner mixed-use districts close to the city centre to residential suburbs, contain major arterial roads other than motorways and contain a mixture of housing stock including traditional high-density tenements, high-rise flats and new housing developments.

FIGURE 4.

Study areas for the core survey. Contains Ordnance Survey data © Crown Copyright and database right 2016.

Documentary analysis

During the pre-construction planning and consultation phase, two key public documents outlined the proposal for the M74 extension and described its potential effects. The first of these, the Environmental Impact Assessment (EIA), was published in December 2003. 61 This document, led by proponents of the scheme within the Scottish Executive, outlined the plans for the new motorway and detailed the hypothesised positive and negative impacts on traffic flows, the environment and the community in the immediate area. After a significant number of objections to the compulsory purchase orders for land required for the scheme, and general public protest, an independent PLI was conducted by a senior planner within the Scottish Executive. The second key document, the report of findings from the inquiry, was published in March 2005. 54

To analyse and compare these two documents, the hypothesised impacts of the new motorway were organised according to the categories defined by the EIA and broadly used by the subsequent PLI (Table 4). As described in Chapter 1, Rationale and approach for the study, the study was limited to examining selected impacts of living near a new urban motorway. Further work on the environmental survey was therefore focused on the categories of impact that were most relevant and feasible for our study:

• land use and landscape appraisal

• visual impacts

• noise and vibration

• pedestrian and other community effects

• economic effects.

Where a distinction was made between impacts predicted in the short term (e.g. 1 year) and in the long term (e.g. 15 years), our documentary analysis focused on the short-term impacts because these were more congruent with the time frame of our study, which examined impacts approximately 2 years after the opening of the new motorway. Although disruption attributable to the construction period was potentially relevant to our study, we had no way of verifying such impacts because we were unable to collect data during construction and, therefore, did not explore this category of impact further in the documentary analysis. Effects on vehicle travellers were important, in that changes in travel times may have influenced travel decisions. However, the measurement of traffic flows was beyond the scope of our study and was therefore not explored further in the documentary analysis. The potential economic effects of the new motorway were mostly explored with reference to the PLI, because this topic was not covered in detail in the EIA.

The two documents examined projected impacts in a number of key local areas in the M74 extension corridor (Figure 5).

FIGURE 5.

Key local areas in the M74 corridor. Contains Ordnance Survey data © Crown Copyright and database right 2016.

Preliminary community and stakeholder engagement

A complementary programme of community and stakeholder engagement ran alongside the study. This was brokered by the Glasgow Centre for Population Health (GCPH) and the Scottish Community Development Centre (SCDC), which are established centres working with community organisations in Glasgow. Between March and May 2013, initial contact was made with 18 key organisations in the M74 corridor, the interests of which may have been affected by the new motorway. These included local housing associations, development trusts, community councils, residents’ associations and public-sector development agencies. Organisations were invited to comment on whether or not any issues associated with the new motorway were of importance to local communities, and, if so, what these issues were. The process and outcomes of subsequent community and stakeholder engagement in the study are described in Chapter 8.

Interviews with key informants

The interviews with key informants were intended to provide an overview of the environmental, economic and social impact of the new motorway. Informants were recruited purposively, based on their involvement with the local groups identified through the preliminary community engagement described in the previous section in combination with snowball sampling. Informants represented a variety of organisations, including local development groups, local community councils, local housing associations, local charities, organisations involved in the planning and development of the motorway and an anti-M74 protest group [Joint Action against M74 (JAM74)].

Between March 2014 and April 2015, information sheets and invitations were either e-mailed or posted to key informants’ organisations and followed up by either e-mail or telephone call. In cases in which an individual informant had already been identified by initial community engagement work, that informant remained the contact for that group. Twenty-five invitations were sent and 12 key informants consented to take part.

Key informant interviews were carried out using a semistructured format, based on a topic guide that could be applied flexibly to informants’ varying roles and levels of involvement with the local area. As key informants held a variety of roles, some were able to give an overview of issues, whereas others addressed local changes in more specific detail. Interviews were conducted in a variety of locations depending on the informants’ preferences, including their homes, places of work, cafés and the offices of the Medical Research Office/Chief Scientist Office (MRC/CSO) Social and Public Health Sciences Unit. All but one were recorded using a digital voice recorder and transcribed verbatim. One participant declined to be recorded; their interview was attended by two researchers, of whom one conducted the interview and the other wrote notes that formed the material for subsequent analysis. These key informant interviews were much more specifically focused on the M74 than the main programme of qualitative research with local residents described in Chapter 7. The topic guide focused on a number of themes, including the impacts of the physical structure of the motorway and associated engineering features, and its wider economic impacts (see Appendix 1).

Walkalong interviews, in which an interviewer and participant conduct their interview while walking through a particular space, were not originally envisaged as a method for interviewing key informants. However, two key informants suggested that they would be better able to address the main themes of their interview by showing their local area as well as describing it. We therefore conducted walkalong interviews with these two participants. More information on this method can be found in Evaluating the intervention.

Field visits and analysis of aerial imagery

In 2013, a member of the study team examined various aspects of the motorway infrastructure using Google Street View. In 2014, another member of the study team undertook several field visits to inspect the motorway and its relationship with the wider cityscape. Examples are shown in Figures 6–9.

FIGURE 6.

M74 extension crossing the Tradeston area on a viaduct. Photograph © Amy Nimegeer and reproduced with permission.

FIGURE 7.

M74 extension crossing the Eglinton area on a viaduct. Photograph © Amy Nimegeer and reproduced with permission.

FIGURE 8.

Noise barriers along the M74 extension at Rutherglen. Photograph © Amy Nimegeer and reproduced with permission.

FIGURE 9.

Underpass beneath the M74 extension at Rutherglen. Photograph © Amy Nimegeer and reproduced with permission.

The new motorway formed one component of a wider strategic initiative to regenerate the local area. We therefore wished to identify other concurrent major changes in the built environment. We developed bespoke software to display side-by-side aerial images of the same location taken at different times, using the Google Earth time slider function. The software allowed the operator to zoom the images and move them in tandem, comparing them in order to identify areas of difference and delineate each area with a polygon. The changes identifiable using this method included the construction or demolition of buildings, and the loss or gain of green space (Figure 10).

FIGURE 10.

Example of environmental change identified by comparing aerial images. (a) 2015; and (b) 2005. Source: Google Earth.

We used this method to identify (but not to characterise) all visible changes occurring between 2005 and 2015 in each of the three study areas and extending to approximately 1 km beyond their boundaries (Figure 11). As expected, large changes had occurred in the South study area during this time, but substantial changes were also identified in the other two areas.

FIGURE 11.

Areas of change within and surrounding the study areas. Contains Ordnance Survey data © Crown Copyright and database right 2016.

Participant sampling

Data collection

Incentives and feedback for participants

For the core survey, participants were entered into a £50 prize draw (at baseline) or received a £5 voucher (at follow-up). At follow-up, those who participated in the objective measurement study received a second £5 voucher, and those who participated in the qualitative study received an additional £10 voucher for each interview conducted.

Derivation of key variables

In this section, we describe the derivation of the key variables used in our quantitative analyses.

Travel behaviour

Core survey

We excluded the travel records of participants who returned a completely blank record, reported not having been at home on the day in question, returned a record so implausible that they appeared to have misunderstood the question or returned non-numeric values (such as ticks) instead of minutes values. However, we retained records in which participants had reported no journeys but had completed other parts of the record (e.g. specifying the day of the week), treating these as a positive indication of ‘no travel’ rather than as missing data on travel behaviour. Participants were instructed to report neither journeys made in the course of work (such as driving a bus or making deliveries), because these were not personal travel, nor those made purely for recreation (such as going for a bike ride) rather than to get from place to place, because recreational physical activity of this kind was captured in the physical activity questionnaire. If such journeys were reported, they were deleted from the travel behaviour record. Time spent using each mode of transport was summed and used to derive the following variables:

• total travel time (minutes/day)

• bus travel time (minutes/day)

• car travel time (minutes/day)

• walking time (minutes/day).

Summary variables were not derived for time spent on the train or using a bicycle, because fewer than 6% of participants at follow-up reported using these modes of transport.

Scottish Household Survey

Data for the period 2009–10 were pooled to represent the pre-intervention condition before the opening of the new motorway, data for 2012–13 were pooled to represent the post-intervention condition and data for 2011 were disregarded owing to the fact that the new motorway was opened midway through that year. Each travel diary recorded in the relevant years was parsed into individual journeys and then into individual journey stages. For example, the sequence ‘walked to bus stop, travelled on bus, and walked to destination’ describes one journey comprising three stages. To maximise the available sample size, we used journey stage as the unit of analysis. Each stage of each journey, regardless of its length or purpose, was classified as ‘active’ if it had been walked or cycled. The origin and destination of each stage was assigned to a Scottish Intermediate Zone following a system that classifies addresses into geographical polygons forming groups of approximately 4000 residents that respect physical boundaries and natural communities, have a regular shape and contain households with similar social characteristics. 75 As with the analysis of accidents described above, an expanded (1000-m) buffer was used to define the three study areas for analysis of the SHS travel diary data, and journey stages were assigned to one of the three study areas if the whole or the majority of the Intermediate Zone polygon containing either the origin or destination fell within the relevant expanded buffer. For the South study area, the River Clyde was used as a natural northern boundary when selecting the Intermediate Zones for analysis because it forms a natural barrier to walking and cycling journeys. This is consistent with the original delineation of the South study area (Figure 12).

FIGURE 12.

Intermediate Zones assigned to expanded study areas for analysis of SHS travel diary data. Contains Ordnance Survey data © Crown Copyright and database right 2016.

Exposure to the intervention

We used multiple ways of describing exposure to the intervention, as we hypothesised that different types of exposure would be more or less important for different outcomes. 80

Areal measure of exposure

As described in Evaluating the intervention, survey participants were sampled from three local study areas: an area surrounding the new M74 motorway (South); an area surrounding the established M8 motorway, which was built in the 1960s (East); and an area containing no comparable motorway infrastructure (North) (see Figure 4).

Individual measures of exposure

Using a GIS, we calculated the distance in metres from the weighted population centroid of the unit postcode for each participant’s home address (1) in a straight line to the nearest motorway infrastructure, and (2) by road network to the nearest motorway junction. Depending on the study area, the nearest motorway may have been the new M74 extension or the existing M8 motorway. We hypothesised that the effect of a unit change in distance would be greater among those living closer to motorway infrastructure. We therefore transformed these individual exposures using the negative natural logarithm to produce a measure of proximity, in which higher values represented greater exposure. We considered the straight-line distance to be the most appropriate exposure measure for the analysis of effects on well-being, because the experience of some of the hypothesised influences of the intervention on well-being (such as noise and vibration) was likely to depend on direct proximity. However, we considered the road network distance to be the most appropriate exposure measure for the analysis of effects on travel behaviour and physical activity, because these were more likely to depend on the routes available for people to follow.

Secondary measures of exposure

As described in Characterising the environmental changes and refining the study design, we used bespoke software to identify changes in the built environment using time-stamped aerial images. Polygons denoting the areas of change were then projected into a GIS, and the area in square metres of each change polygon was calculated. Using the GIS, we constructed an 800-m pedestrian network buffer around the weighted population centroid of the unit postcode for each participant’s home address, and identified the proportion of the area within that buffer that had changed. This gave an indication of the general amount of change occurring in the neighbourhood (an ‘upheaval index’), whereby higher values represented a greater amount of change (Figure 13). Examples of neighbourhoods experiencing different amounts of change can be found in Figure 14.

FIGURE 13.

Areas of change within participants’ neighbourhoods. Contains Ordnance Survey data © Crown Copyright and database right 2016.

FIGURE 14.

Examples of neighbourhoods experiencing different levels of upheaval. (a) 1%; (b) 5%; and (c) 10% of the local built environment changed between 2005 and 2015. Contains Ordnance Survey data © Crown Copyright and database right 2016.

Overall approach to analysis

STATS19

A total of 78,919 road traffic accidents were recorded in Glasgow City and surrounding authorities between 1997 and 2014. A total of 13,595 of these resulted in serious injury or death. The total numbers for the South, East and North study areas were 10,167, 3832 and 3956, respectively. The annual count fell in each study area from 1997 to 2014. For all accidents, the annual count fell by 50.7% in the South (from 758 to 374), by 49.3% in the East (from 292 to 148) and by 50.5% in the North (from 315 to 156). For serious and fatal accidents, the annual count fell by 57.4% in the South (from 195 to 55), by 71.4% in the East (from 70 to 20) and by 68.6% in the North (from 70 to 22) (Table 10).

The annual numbers of casualties are shown in Table 11. The numbers of casualties do not match the numbers of accidents because some accidents result in multiple casualties. As with accidents, the annual numbers of casualties of all types fell from 1997 to 2014 in the region as a whole, and in each of the three study areas.

Scottish Household Survey

A total of 3706 and 4205 individual travel diaries were completed by SHS participants in Glasgow City and surrounding local authorities during the periods 2009–10 and 2012–13, respectively. The numbers of individual journey stages recorded were 9777 (2009–10) and 11,684 (2012–13), respectively. Each participant was assigned to a study area based on the Scottish Intermediate Zone containing their home address (see Chapter 2, Derivation of key variables). The numbers of journey stages recorded in the South, East and North study areas were 434, 477 and 541, respectively in 2009–10, and these increased to 543, 560 and 593, respectively, in 2012–13.

Descriptive characteristics of the SHS travel diary sample can be found in Table 12. At both time points, there were more women than men, the 45- to-59-year age group contributed more participants than any other and most participants were working or studying.

Patterns of travel behaviour

Using the criteria described in Chapter 2, Evaluating the intervention, 1141 and 1206 travel records were found to be suitable for analysis at baseline and follow-up, respectively. A total of 71 and 92 ineligible journeys were removed at baseline and follow-up, respectively.

Patterns of travel behaviour in the longitudinal cohort and repeat cross-sectional sample can be found in Table 13. At both time points, car and walking were the most frequently reported modes of transport, with approximately half of the sample reporting using them at T1, and one-third reporting using them at T2. The proportions of participants reporting any travel, and using any particular mode of transport, were, on average, significantly lower at follow-up than at baseline. For example, 91% and 85% of longitudinal and repeat cross-sectional participants, respectively, reported any travel at T1, and these proportions fell to 69% and 65%, respectively, at T2. Among those who travelled, however, the average daily time spent travelling or using a particular mode of transport was fairly consistent over time.

Table 14 presents unadjusted summary measures of travel behaviour by study area. It shows few large differences between areas, apart from a higher likelihood of using the bus among cohort study participants living in the East.

Patterns of physical activity and sedentary behaviour

Using the criteria described in Chapter 2, Evaluating the intervention, 958 and 1029 records were found to be suitable for the analysis of MVPA, 1050 and 1095 records for the analysis of walking, and 917 and 984 records for the analysis of sitting time, at baseline and follow-up, respectively.

Patterns of physical activity and sedentary behaviour in the longitudinal cohort and repeat cross-sectional sample can be found in Table 15. The proportions of participants reporting walking and MVPA were fairly consistent between time points, with > 80% of the sample at each time point reporting some walking and more than two-thirds reporting some MVPA. Similarly, self-reported sedentary behaviour was fairly consistent over time, at an average of approximately 380 minutes (6.3 hours) per day. In the repeat cross-sectional sample, however, among those who reported any walking, the average time spent doing so was lower at follow-up than at baseline (355 vs. 410 minutes/week; p = 0.008).

Table 16 presents unadjusted summary measures of physical activity and sedentary behaviour by study area. It shows no large differences between areas at either time point.

Relationships with environmental exposures

Patterns of travel behaviour

The proportions of journey stages made by active modes of transport are shown in Table 21. All areas experienced small increases in this proportion from 2009–10 to 2012–13 (from approximately 21% to 24%) and there was little variation between areas in this respect. The wider region of Glasgow and surrounding authorities had a smaller proportion of active journey stages in 2009–10, and also experienced a smaller increase over time compared with the South, East and North study areas. This may reflect the much larger (and in parts rural) area covered by the wider region.

Relationships with environmental exposures

Patterns of physical activity and sedentary behaviour

Using the criteria described in Chapter 2, Evaluating the intervention, 196 records were suitable for analysis. However, three participants, despite providing some GPS data, did not provide adequate data to allow for the derivation of summary measures of activity in their neighbourhood, which reduced the sample available for those analyses to 193.

Patterns of physical activity and sedentary behaviour are summarised in Table 24. Participants included in these analyses provided nearly 7 days of data on average, wearing their accelerometers for an average of 861 minutes (approximately 14 hours) per day and recording a valid location on their GPS receiver for an average of 561 minutes (approximately 9 hours) per day. Most participants exceeded the minimum recommended daily quantity of MVPA, with an average of 45 minutes per day in the sample as a whole. About half of all light physical activity, and one-third of MVPA, occurred within the 800-m pedestrian network buffer used to define each participant’s home neighbourhood.

There were no large differences in unadjusted summary measures of physical activity or sedentary behaviour between study areas (Table 25).

Relationships with environmental exposures

Analytical strategy

Interrupted time series regression models with ARIMA errors were fitted to the monthly accident count data to test whether or not there had been a change in the secular trends of accidents during the construction of the new motorway or following its opening. 97,98 Individual models were fitted to each study area and data series. Log transformations and differencing were applied to the data to achieve time series that were normally distributed and stationary in both level and variance. For series in which there were months with a zero count of accidents, such as those limited to serious and fatal accidents in a single study area, the series were transformed using an inverse hyperbolic sine function. 99 To take account of the non-independent nature of the series, autocorrelation functions and partial autocorrelation functions of each model were used to identify seasonality and guide the initial model building. Detailed residual diagnostics were used to obtain a model with more accurate coefficient estimates. Outliers were identified following visual inspection of the initial models (p < 0.05); dummy variables for these outliers were then included and the models were re-estimated. The Akaike information criterion (AIC) was used to broadly assess model fit and to guide choices between competing models, with a lower AIC value indicating an improved fit. 100 The AIC was also used to assess whether or not removing outliers improved overall fit.

Results

Table 28 presents the results from the ITS models for (1) all accidents and (2) serious and fatal accidents only. There was a significant downwards trend in the monthly count of all accidents in each of the study areas over the period 1997–2014.

Patterns of well-being

Patterns of mental and physical well-being in the longitudinal cohort, the repeat cross-sectional sample and the full follow-up sample can be found in Table 30. In all samples, and at all time points, average well-being measured using the SF-8 was lower than in the general US population in which the instrument was developed, which has mean PCS-8 and MCS-8 scores of 49. 66 The Warwick–Edinburgh Mental Well-being Scale was administered in the 2012 Scottish Health Survey and a mean score of 50 was reported,118 but comparable population estimates for the short version are not available. In both longitudinal and repeat cross-sectional samples, physical (but not mental) well-being was significantly lower at follow-up (when participants were older on average) than at baseline.

Table 31 presents unadjusted summary measures of well-being by study area. There were no large differences between study areas at either time point.

Relationships with environmental exposures

Analytical strategy

We calculated descriptive statistics for neighbourhood perceptions at follow-up using the full T2 sample. We used t-tests to test for differences between study areas. We then calculated descriptive statistics for change over time in perceptions of the physical neighbourhood (both individual items and the factors derived from them) in the longitudinal cohort.

We undertook two sets of analysis, both limited to participants living in the South and East study areas (the areas with a motorway). The first set of analyses used the full sample recruited from the South and East study areas at follow-up (n = 750). Linear regression analyses, adjusted for the same set of covariates as in the main analyses, were used to test the cross-sectional associations of proximity to a motorway with neighbourhood perceptions and collective efficacy, stratified by study area.

The second set of analyses used the cohort sample living in the South and East study areas (n = 209). Mediation analyses were conducted following the method of Baron and Kenny86 (Figure 18), adjusted for the same set of covariates as the main analyses and stratified by study area. The exposure (X) was proximity to a motorway. Mediators (M) were changes in perceptions of the neighbourhood (either items or the factors derived from them). The outcome (Y) was change in mental well-being (MCS-8). We conducted single mediator models, with a Sobel–Goodman test to assess the significance of the mediation effect.

FIGURE 18.

Baron and Kenny mediation analysis. Adapted from Baron R, Kenny D. The moderator–mediator distinction in social psychological research: conceptual, strategic, and statistical considerations, J Personal Social Psychol, 51, 1173–82, 1986, published by the American Psychological Association. 86

Results

The M74 extension and associated physical changes

Participants discussed a number of features related to the M74 extension including the carriageway itself, its viaducts, underpasses, embankments, slip roads, junctions and planted verges, and its integrated pedestrian infrastructure. The physical structure itself evoked mixed feelings:

So, yeah, so from that perspective the M7 – I’m pro the M74 is what I’m trying to say . . . I think like whereas before you just had a big patch of wasteland, now you’ve got a nice shimmering motorway.

Man aged 20–35 years living in Govanhill

It’s less aesthetically pleasing.

Key informant, local advocacy group

Some participants drew attention to the imposing nature of the structure, with some describing it as either ‘impressive’ or ‘ugly’, whereas others felt that they had grown so accustomed to it that it now seemed as if it had ‘always been there’. For one respondent, this familiarisation was attributed to the long construction period:

It took such a long time to build it that it just gradually came, so it didn’t just suddenly appear overnight, it took time, so you’re just, just used to it now.

Man aged 36–50 years living in Rutherglen

For another respondent, it reflected to the large amount of change to which local residents had become accustomed:

I think folk are just used tae seeing huge swathes o’ . . . there’s no’ that sense and nobody complains aboot [change], you know? I just think people have got used tae change happening.

Key informant, local community association

For the most part, participants considered the underpasses to be well thought out, open and relatively well lit:

[It is l]it up an’ things like that so it’s . . . quite safe-looking. Like, you wouldn’t feel paranoid walking under it an’ that sort o’ thing so it’s . . . the motorway’s not really affected access to things as well. They’ve thought it out really well.

Woman aged 20–35 years living in Rutherglen

However, one participant living in Rutherglen felt that her new local underpass made her evening commute marginally darker.

Some participants praised the efforts to plant greenery on the motorway embankments:

I think there was an improvement because the landscaping and bits and pieces that they had done did certainly improve the outlook – and the new bridges and the work they done on the bridges kinda did actually genuinely brighten up that area. And most of the land, as I says, having been going in and out of Govanhill from kinda early seventies, most of the land was lying derelict, anyway.

Man aged 51–65 years living in Govanhill

However, others were less convinced about the choice of planting, given that broad-leaved trees shed their leaves for much of the year:

Well the screening planting that I’ve seen isn’t dense. It’s more of a . . . aesthetic planting than screening ‘cos it’s not anywhere near dense enough and layered enough to perform anything that you would want from screening and also the species choice. I mean if you choose broad-leaved trees they drop their leaves in the winter. Actually it’s the winter when the motorway’s loudest because that’s when the road is wettest and tyres on wet tarmac make a lot more noise than tyres on dry tarmac.

Key informant, local greenspace organisation

In addition, some key informants questioned who was responsible for the upkeep of these planted areas.

Other physical elements of the structure that were discussed by both residents and key informants included slip roads and junctions. For the most part, these were mentioned in the context of problematic traffic flow, which was thought to be related to poor design or planning and was particularly troublesome around peak commuting times. One junction near Govanhill in particular was mentioned several times as being problematic. New slip roads were also identified as challenging where participants considered that lanes ended in an abrupt and unsafe manner:

They haven’t designed the slip roads properly, in a sense that you’ve kind of going . . . if you want to kind of go straight on, if you’re coming in this direction here and you want to go straight on, they’ve kind of designed the slip road onto the M74 as you kinda got to kick and get onto it . . . what that’s meant is that three lanes are filtered into two lanes and that causes tailbacks either way . . . you often see a snarl up of traffic there and a snarl up of traffic there.

Key informant, local housing association

Most of the new motorway is elevated, but at one point near Govanhill it passes beneath a pre-existing arterial road. This creates very different effects at different places (Figures 21 and 22).

FIGURE 21.

(a) Elevated sections of the M74 extension; (b) elevated sections of the M74 extension close to housing in Laurieston; and (c) elevated sections of the M74 extension in Rutherglen. Photographs © Amy Nimegeer and reproduced with permission.

FIGURE 22.

M74 extension passing beneath an existing street in Govanhill. Photograph © study participant and reproduced with permission.

One particularly successful and much praised aspect of the new infrastructure was a new pedestrian overpass in Govanhill, which was considered by all who commented on it to have been well thought out and to represent a significant improvement over the existing pedestrian bridges over the railway line. Cycling infrastructure, however, was not as highly praised (for more detail, see Active travel).

Other local streets

The M74 extension did not occur in a ‘vacuum’. Within the time period of interest to the study, residents reported changes to the design and surface of local streets (e.g. widening of pavements, appearance of pot holes or resurfacing), traffic calming measures and other, ongoing, local transport projects. For example, a one-way system was implemented on some residential streets in Govanhill. Some regarded this as funnelling traffic onto Aikenhead Road (a main arterial road) and causing tailbacks, whereas others saw it as making local streets safer for children to play in. More than one participant discussed changes to the layout of Aikenhead Road, citing it as a possible contributor to poorer traffic flow and increased stress for local drivers (Figure 23). Such concurrent structural changes made disentangling the impacts of the M74 extension a challenge not only for researchers, but also for local residents.

FIGURE 23.

Reconfiguration of Aikenhead Road. (a) In 2015 (red line); and (b) in 2005. Source: Google Earth.

Local amenities

Residents and key informants discussed changes to local amenities including shops, schools, health facilities and green spaces. These were considered by residents to be important spaces for connecting with other local people (both casually and intentionally) and were central to conceptualisations of their neighbourhoods as healthy or desirable spaces. Shops in particular emerged as important local meeting places, imbued with the additional symbolic value of representing the character of the local area. In both Govanhill and Rutherglen, participants described concerns about a decline in the quality of their local high-street shops that was unrelated to the opening of the new motorway. In Rutherglen, several residents expressed concern that what had once been a unique, vibrant shopping area that reflected the town’s distinctive status as a royal burgh had been overtaken in recent years by betting shops, pound shops and charity shops (Figure 24):

The shopping area, the Main Street in Rutherglen has deteriorated . . . I mean, when you look at it an’ it’s all the bookies and the . . . I think there’s a pawn shop. Of course, I mean, they’re supplying a need . . . but . . . that is to the detriment of people.

Woman aged 65+ years living in Rutherglen

FIGURE 24.

Changing high-street shops in Rutherglen. Photograph © study participant and reproduced with permission.

Several residents expressed concern that the changing shops broadcast an unfavourable presentation of Rutherglen to outsiders as a community in decline. Similarly in Govanhill, several residents mentioned concerns about declining high-street shops, but with a slightly different emphasis reflecting recent demographic changes. A number of local shops now specialised in ‘ethnic’ clothing and foodstuffs. For some, this change in the nature of the local shops was indicative of a decline in traditional local culture:

[I]t’s not nice seeing so many closed shops, or shops being taken over by ethnic groups and things, you know?

Woman aged 51–65 years living in Govanhill

However, for others it was something to be celebrated, giving Govanhill a unique and multicultural character:

It’s a very diverse area. There’s quite a high, a high population of immigrants there. But that’s kind of translated into lots of interesting shops and things like that.

Man aged 20–35 years living in Govanhill

Although these changes to local shopping areas caused consternation, they were attributed by participants not to the opening of the new motorway, but rather to wider economic or social factors including rising car use. Some Govanhill residents had initially worried that the new motorway would cause people to bypass the local shopping area centred on Victoria Road. When this did not come to pass, however, they made their peace with the motorway and its presence:

I thought [the motorway] would destroy what’s left of the shops in Govanhill. I thought they were, be gone . . . so, I was very strongly against the motorway being built . . .

What do you think it was that kind of changed your mind about like . . .?

I think that it was, we didn’t become a ghost town and I think that’s the reason why.

Man aged 36–50 years living in Govanhill

One significant change for both communities was the opening of new ‘big box’ supermarkets nearby. Given the importance placed by residents on their local shops, it was perhaps surprising that these supermarkets were almost universally praised as beneficial to the local area, and were also described as frequent destinations for active travel (mirroring the findings of the baseline study57). One participant also described how cashiers in their local supermarket took care to remember the names of customers, particularly those perceived as being lonely and in need of connection.

Public transport

As might be expected in areas of comparatively low car ownership, public transport was important for residents of both areas. For the most part, residents considered both Govanhill and Rutherglen to be very well served by bus and rail services. However, recent changes to bus timetables and to the locations of bus stops had caused difficulties for some people with impaired mobility:

There used to be a bus stop in the middle and they seem to have taken it away at this side walking down towards [neighbouring road]. There is one in the middle at the opposite side but it is really quite a long walk. There’s no bus stop. I don’t mind walking but it’s for likes o’ the elderly, you know.

Woman aged 51–65 years living in Rutherglen

They moved bus stops and it’s – it passes the 101. Used to stop outside the bookies but now they’ve taken it up to the cash and carry so it’s quite a wee walk back for me and it’s a bad pavement and a bad road and a big kerb, so that’s awkward for me.

Woman aged 65+ years living in Govanhill

Particularly in Govanhill, these changes were described as having led to more hazardous walks to the bus stop, and longer and more convoluted journeys. In Rutherglen, however, refurbishment of the railway station was viewed very positively by local people. Although residents did not associate these changes to public transport with the opening of the new motorway, for key informants the new motorway was seen as part of a wider investment programme of which the causes and effects were difficult to disentangle. For example, Clyde Gateway – a regeneration company tasked with the physical, social and economic development of the east end of Glasgow – was inextricably linked with the new motorway in the minds of some key informants, who therefore thought that some other regeneration activities such as refurbishment of railway stations would not have occurred without the impetus of the investment in the M74 extension. For many local residents, however, investment in some local refurbishments was regarded as not necessarily for their benefit. On the contrary, many of them considered these changes to have been made for the benefit of people attending the Commonwealth Games in 2014, to present an attractive version of Glasgow to the wider world:

Yeah, just as well we had the Commonwealth Games or it would have still have been the way it was, you know? There’s no way in God’s creation, would I use that railway station, you know, but I would use it now if I had to, you know? But I wouldn’t have used it before, because it was too lonely and too isolated, you know? It was quite scary. But they’ve got it really nice, really. They’ve made a big difference to the place.

Woman aged 51–65 years living in Rutherglen

One of the things we got done was [the main road] outside my house got resurfaced. This is great, and like everything like that was getting done with the Commonwealth Games, you know, people were being so negative. I’m like, ‘Why are you being negative? We’re getting the road resurfaced. And it’s great’.

Woman aged 36–50 years living in Govanhill

These two quotations illustrate how local people may have benefited from changes even if those were perceived to have been mainly intended to benefit others. Other material suggests differences between those who benefited from changes and those who incurred the harms, raising questions of social justice (see Severance and connectivity).

Aesthetics

Other physical changes described by local residents included the demolition and construction of residential and commercial properties. Although no single building stood out in this respect, discussion of the razing of older buildings did draw attention to the rich historical background of the two case study communities, which was often prized by older residents and seemed to be central to their feelings of connection to place. However, residents also described a decline in the cleanliness and upkeep of buildings and street areas [including common spaces, closes (communal entrances), facades and front gardens]. For residents, this was most commonly associated with changing local demographics. The in-migration of people perceived as not sharing tacit local values relating to property upkeep, owing to either cultural differences or delinquency, and the rise of absentee landlords renting properties on a short-term basis to people with no local roots or connections, were frequently cited as causes of the decline in local aesthetic upkeep:

[. . .] the appearance of certain streets that’s really – you walk along and, you know, there’s rubbish out the front and there’s, you know, old prams that are just lying in the street or shopping trolleys, you know . . . I don’t remember it being, you know, as bad to be honest . . . you know, as I say about 10 years ago, probably about 10 or 11, maybe 12 – that’s when I did start realising you can see like, you know, certain windows ‘cos you would look and go that’s not curtains in the window, that’s linen sheet up there, you know, and the colour of windows, you know, and when the house looks manky, that close looks terrible, you know?

Woman aged 36–50 years living in Govanhill

Traffic

For many local residents, a significant factor influencing their feelings towards the new motorway was the extent to which they believed that it had affected levels of traffic on their local streets, or the extent to which their perceptions of these changes matched their prior expectations. Perceptions of whether or not traffic on local streets had changed were mixed:

Well I thought it would make a big difference but . . . I don’t see a great deal o’ difference because the traffic’s still just much the same oot here as it always was.

Man aged 65+ years living in Govanhill

It’s reduced the amount of traffic or lorries, big traffic, big trucks an’ things like that going along the Main Street. It’s certainly reduced all that.

Man aged 51–65 years living in Rutherglen

At the first we said, ‘well look, it’s taken a’ the traffic away, that’s great’. But they’ve sneaked back in again, you know, withoot anybody noticing. But it was a good thing. We were always for it. Away back when we stayed in wur [our] other hoose you got a lot of campaigners saying ‘no’ to this. I said, ‘Aye, I want it’. I wanted it a’ [all] the time, you know?

Man aged 51–65 years living in Rutherglen

On the whole, residents in Rutherglen were more likely than their Govanhill counterparts to report a definite decrease in traffic on local main roads. This conforms with the expectations set out in the EIA for the new motorway. 61 Residents in Govanhill were more likely to report that traffic had remained the same or increased, particularly those living on main roads, or to say that they were not sure. Key informants consistently identified London Road – a main arterial road, outside the qualitative case study neighbourhoods but linked with the new motorway – as one on which traffic had significantly decreased:

2 or 3 years ago before the extension was built it was really sometimes quite tricky even to, as a pedestrian, to get across London Road from one side to the other.

Key informant, local housing association

As mentioned above, however, residents attributed changes in traffic levels to different causes, for example:

And do you think that, since the motorway has come in, have you noticed any change in your local traffic?

No. No. As I said, they’ve changed the local traffic ’cos they’ve messed about wi’ all the roads an’ blocked off streets an’ shoving bus lanes in an’ . . .

So other changes have had more of an impact than that?

Yeah, definitely.

Man aged 51–65 years living in Govanhill

Noise and air pollution

Two key elements of experiential change described by participants were those of noise and pollution. Perhaps unsurprisingly, these were heavily intertwined with discussions of traffic and the M74 extension, and perceptions of noise and air pollution were related to perceptions of local traffic levels and the proximity of people’s dwellings to the motorway. In most cases, participants who perceived local traffic to have decreased also considered their exposure to noise to have decreased. However, a handful of participants acknowledged a decrease in local traffic but still believed that their exposure to pollution may have increased because of their proximity to the new motorway. A number of residents and key informants drew particular attention to the possibility that people living in extreme proximity to the motorway, for example in Devon Street or close to the junction with the M8 motorway, might experience particularly severe impacts. Discussions about pollution were manifested in three main ways. The first involved descriptions of an increase in settled particulates or grime on windows or in houses, which was quite commonly described in both Govanhill and Rutherglen:

[. . .] you have tae wash your windows a little more frequently ‘cos o’ the fumes and what-have-you.

Man aged 51–65 years living in Rutherglen

Where we are, now we run along [a main road] which we have noticed the smog, and the dirt and the dust, impossible to keep windows clean ‘cos the back of our flat goes onto [the main road] so you’ve got all the motor traffic there. So that’s quite difficult to cope with at times.

Woman aged 36–50 years living in Govanhill

The second, less frequent way involved health effects such as asthma and wheezing:

So if you open your windows you can notice it?

Yeah. Mm hmm.

OK. And you can – what can you smell it, or . . .?

It’s no’ smelling like, like congested, I speak (wheeze) like – [gestures to her chest]

In your chest?

Mm hmm.

Woman aged 36–50 years living in Govanhill

The third involved fumes:

So how has that changed that walk for you?

It just makes it a little bit more noisier and, obviously, pollution from the petrol fumes o’ the thingmy makes it a little bit . . . well, that’s changed it. There’s more pollution and things like that as well.

Man aged 51–65 years living in Rutherglen

For participants, therefore, pollution perceived to be attributable to the new motorway was something that could affect them both inside and outside their homes. During the course of the qualitative fieldwork, Friends of the Earth (London, UK) published an analysis of pollution in 2014, which identified Rutherglen High Street as one of 13 sites in Scotland to exceed the European Union’s legal limit for nitrogen dioxide and also reported that particulate matter of a diameter of ≤ 10 microns had increased between 2013 and 2014. 121 This story received news coverage, and a handful of participants were aware of it (see Chapter 2, Characterising the environmental changes and refining the study design):

[. . .] you’ll know that Rutherglen Main Street has some of the worst air quality in Scotland. There’s been quite a lot of press reporting on that recently.

Key informant, local community council

It took away some of the traffic but not all of it, so the Main Street’s still very heavily used, you see cars backed up all the time. So, I guess that causes a lot of pollution. In fact I’m sure it was Rutherglen Main Street came up pretty bad recently in a pollution survey.

Woman aged 36–50 years living in Rutherglen

Both residents and key informants referred to ground contamination in both Rutherglen and Govanhill, a legacy of their heavy industrial history. For some participants, their greatest concern about pollution from the motorway was that the construction might have disturbed land contaminated with chromium from former chemical works:

But they obviously were aware of the toxic waste . . . It’s just under [former site of works] and there is . . . well . . . I mean, it’s too late now. I mean, the M74’s there. My concern and always has been is with the dust, prevailing wind . . . driving it over.

Woman aged 65+ years living in Rutherglen

For more than one key informant, however, building the new motorway on contaminated land was a way of making the land useful again by ‘capping’ the contaminants, and the project provided an impetus to decontaminate sites in a way that would not otherwise have been economically viable. It should be noted that the chromium contamination of land in Rutherglen had been the subject of a health impact assessment before the motorway was built and was, in fact, the main health impact considered therein. 122 This pre-existing, historically rooted understanding of the local communities as post-industrial ‘polluted places’ may have coloured local perceptions of the impacts of the new motorway.

People’s experiences of noise were also variable. For some, the new motorway had led to a perceptible increase in noise, even intruding on their enjoyment of their outdoor spaces and reducing their desire to open windows at home, whereas others felt that it had had little perceptible impact and were in fact impressed at the low levels of noise they experienced. These differences were not always attributable to how close people lived to the new motorway, and may also have been related to individual habits and sensitivity to noise:

And it is quite noisy if you . . . open a window and it’s, you can hear it, and you if you go out in the garden before that came in it was really, really peaceful here . . . occasionally a train would pass but even then it was so low down there are so many trees. But the motorway’s a different thing. The noise of a motorway is really loud, isn’t it? And it seemed to just be humming in your head all the time, you’re sitting like that, and you go, ‘God, I wish that would stop’.

Woman aged 65+ years living in Govanhill

No. I can’t [hear the traffic from here], no. In all honesty, I can’t. But maybe people who stay [live] . . . there’s [street] there. They may hear it because they’re nearer. Where I am, I mean, it used tae be you could . . . even the railway, you could only hear the, or feel the slight . . . not a lot of tremor but you knew the railway was there. But you can’t hear the traffic.

Woman aged 65+ years living in Rutherglen

Some participants took pains to describe the difference between the noise from the existing railway line and the noise from the motorway. For some, the noise from the railway line was greater and, being intermittent, was harder to ignore:

It’s strange, because the railway line’s there, we used to notice vibrations when the heavy freight went by. You know, we had a couple o’ display cabinets and if they weren’t set up properly, you could hear the glass kinda chinking inside – but we never noticed that. It wasn’t, the lorries or anything like that, weren’t aware of anything like that wae [with] the vehicles.

Man aged 51–65 years living in Govanhill

Traffic noise was also discussed in relation to green space, which is discussed at greater length below (see Severance and connectivity).

Active travel

In the qualitative study participants described both positive and negative aspects of their experience of active travel, some of which related to the motorway and some of which related to other factors including health, the weather and the seasons.

As previously mentioned, several residents drew attention to integrated pedestrian infrastructure that was built as part of the M74 extension. For the most part, this was characterised in positive ways. In areas where the motorway was elevated, underpass areas were considered by many to be well-lit, wide (therefore allowing people to pass without invading each other’s personal space) and relatively secure. In particular, participants praised a new pedestrian overbridge in Govanhill. This replaced a previous crossing over the railway line, which was considered to be dark, narrow and unsafe to the extent that participants actively avoided using it. In contrast, the new pedestrian overbridge was described as open, light and secure, contrary to the fears expressed by some participants in the baseline interview study that such crossings would be more frightening and might deter walkers. 58 However, cycling infrastructure around the M74 extension was less widely appreciated. It was described as disjointed, with cycle lanes that began and ended abruptly, requiring cyclists to enter and exit the traffic on main roads in a way that felt unsafe:

Also the access at one of the junctions – well, certainly I don’t know about all of the junctions but certainly one of the junctions – yes, junction 2, it has – when the whole road system was changed with the slip roads and so on, provision for cycling was incorporated into that. But it’s one of these classic examples where you come up to it, the cycle infrastructure starts, you get up beyond the junction and the cycle infrastructure stops. So it’s – it’s good in the sense that it made provision. Because if you, you know, build these things to start with it’s better than retro-fitting them.

Key informant, local community council

Similarly, changes in traffic flows had changed people’s experiences of cycling in both Govanhill and Rutherglen:

So [junction] is the main junction that I would use. As a cyclist going through there, the cycle routes kind of cross over each other, so that’s . . . and you can’t get through, where there used to be a bike lane you can’t get through, where there used to be a bike lane there isn’t one now. So yeah, that’s changed and that’s definitely more dangerous ‘cos the density of traffic has increased.

Key informant, local greenspace organisation

In key informant interviews, it was noted that residents in one community to the north of the motorway expressed concern about the safety of their children walking to school, owing to an increase in traffic around the motorway slip roads. In Rutherglen, however, a decrease in traffic on the main street was perceived to have had the unexpected side effect of more cars being parked on the street. Together with existing cycling infrastructure that was regarded as confusing, in that it comprised a mixture of dedicated and shared-use routes, this constituted a worsening of the obstacles to cycling, which were perceived as a potential deterrent for inexperienced cyclists:

It makes it difficult for [new cyclists]. It’s not smooth to get from A to B. You’re getting off your bike, you’re walking it across. That’s if you’re doing what the sign tells you to do. On this stretch here that’s picture three, where I had the altercation. ‘Do you know . . .’ . . . someone shouting at me, ‘Do you know it’s illegal to cycle on the pavement?’ ‘Well, yeah I do, but not when it’s shared use it’s not’. I know that. But somebody that’s just started cycling might be a bit more intimidated about it.

Woman aged 36–50 years living in Rutherglen

In other words, the impacts of the new motorway combined with other factors in sometimes unexpected ways to produce changes in the local experience of active travel.

Some participants who perceived local traffic to have decreased (whether or not this was held to be associated with the new motorway) considered this to have benefited local people’s capacity and inclination to use active modes of transport:

I think the motorway certainly would’ve changed the character because we don’t get as much through traffic now, as you’ll see from that. And that’s changed the character, if you like, and it’s much easier tae get out and about and crossing roads and things like that. Much, much easier.

Man aged 51–65 years living in Rutherglen

For others, however, the presence of the motorway – despite being unlikely to change, deter or reroute active journeys – was described as affecting the quality of the active travel experience:

It’s a pretty busy road a’ [all] the way out that, depending on the time o’ day again. But it wouldn’t stop me. It wouldn’t stop me. Now, going that way, you’re going away from it . . .

So it’s more just that it’s changed the, sort of, experience of walking?

Yeah, it’s really . . . aye, visually it’s, you can see it, you’re aware of it more. Although sometimes it’s subconsciously, you’re no’ even thinking aboot it, but you know it’s there. You know it’s there.

Man aged 51–65 years living in Rutherglen

Such change in the quality of the experience may have the potential to erode the likelihood that people would make active journeys in future when an alternative was available.

Many participants cited work and other ‘utility’ journeys as their main reasons for active travel. Local high-street shops, ‘big box’ supermarkets nearby and green spaces emerged as important destinations. Beyond their immediate areas, some people also described recently developed pedestrian infrastructure, for example on the Clyde Walkway and in Glasgow Green, as places they enjoyed frequenting.

Aside from any effects attributable to the motorway, one factor emerged as being of major importance to local residents in determining active travel: their perceptions of personal safety. This was mentioned by people living in Rutherglen, who referred to crime related to drug and alcohol use and to an existing pedestrian underpass that was considered unsafe (Figure 25), which was also mentioned by participants in the baseline study,58 but it was a more dominant concern in Govanhill.

FIGURE 25.

View inside an existing pedestrian underpass in Rutherglen. Photograph © study participant; reproduced with permission.

Numerous participants from Govanhill described a decline in their perceived personal safety over the study period, which they associated mainly with incomers to the local area, along with changes in patterns of tenancy (see Aesthetics and the built environment) and, to a lesser extent, with drugs, alcohol and unoccupied young people:

I couldn’t exactly say when I did notice that ‘cos there’s certain, you know, streets I didn’t walk down just for the fact I didn’t have to walk down thae [those] streets. And it wasn’t because I went ‘I’m not walking down there’, just because I didn’t, you know, to get there I knew well, I can go that way, that’s quicker going, you know, cutting down that street . . . But one time I didn’t give it a thought walking down places whereas the last – definitely the last 5 years I just went ‘That’s it, I’m going in the car’.

Woman aged 36–50 years living in Govanhill

This change in perceived safety led the participant in question to avoid particular streets, and to make more journeys by car that she would previously have walked. Another resident in Govanhill described how she would rather have her children play video games at home, where they were safe, than allow them to play outside unsupervised:

I don’t ask my children [to] go outside because I’m scared.

Woman aged 36–50 years living in Govanhill

These perceptions of declining neighbourhood safety, and of the causes of that decline, had implications for community cohesion and the potential for local people to form neighbourhood relationships (see Severance and connectivity).

Sheltered housing event

This event took place in the common room of the housing complex and was attended by five residents. Two had homes that faced directly onto the new motorway; these were the only residents present who had lived in the complex before, during and after the construction of the new motorway. The other three had homes facing away from the motorway and had lived there only after the motorway was opened. The room was set up with a large map of the area on a table, and sticky notes and pens for annotation.

The two residents with homes facing the motorway highlighted stark problems with noise, vibration and air pollution that began during the construction period and continued after the motorway was opened. They described having to constantly clean their windows and spectacles because of the build-up of dirt, suffering from ongoing eye problems and being unable to open their windows because of the noise and air pollution. They also described sleep deprivation owing to motorway maintenance activities occurring at night. Although the three residents in homes facing away from the motorway also identified local problems with air and noise pollution, they were less sure that the new motorway was the main cause of these and were either indifferent or positive towards it overall. All the residents described increases in local traffic and in heavy goods vehicles using local streets; this was attributed to the motorway as well as to other changes in street layouts and the locations of bus stops. One resident who owned a car felt that the motorway helped her to get around, and others who did not own cars described the motorway as enabling family members to visit them more easily. Residents described difficulties in walking on uneven pavements and a lack of local amenities within walking distance, as well as changes to bus routes that paradoxically made it easier to reach amenities further away than to reach those nearby. Finally, although residents could highlight local road traffic accident hotspots, none felt that the motorway had affected the frequency of accidents.

Rutherglen event

This event took place in a busy shopping centre, frequented by local people as well as those from surrounding neighbourhoods. We set up a banner describing the purpose of the study and directing passers-by to the researchers if they were interested in talking about it. A large map of the area was set up on a table, with sticky notes and pens available for annotation (Figure 28). In total, there were 18 visits to the stand over several hours.

FIGURE 28.

Community engagement event in Rutherglen. Photograph © Amy Nimegeer; reproduced with permission.

Overall, attendees expressed positive, negative and neutral views regarding the motorway. The most consistently expressed view was that the motorway had significantly improved conditions for car users, allowing them faster and more convenient access to surrounding areas and amenities. As at the first event, some residents raised issues of environmental blight relating to noise and vibration from the motorway: one described being unable to open her back window because of the noise, and another described cracks forming in houses nearby. There were conflicting views about whether traffic on Rutherglen Main Street had increased or decreased, but some people felt that a reduction in traffic congestion had improved the local environment. Other comments received included observations that cars and lorries were infringing on cycle lanes on local streets and that public transport provision in the area had decreased, but it was not clear whether or not these were thought to be attributable to the M74 extension.

Gorbals event

This event took place in a community hall, as part of a local lunchtime event, which provided activities for adults and children and a free lunch. We set up a stand comprising the banner, map, sticky notes and pens as described above, and spoke with nine people who lived or worked in the Gorbals.

As at the second event, attendees expressed positive, negative and indifferent views about the motorway. However, unlike at the second event, where discussion focused on the benefits for car users, the discussion at this third event particularly highlighted health and social harms, with four of the attendees having been involved in protests against the plan to build the new motorway. Two mentioned that the PLI had strongly recommended against building the new motorway, but this advice had been rejected and construction had proceeded regardless. The motorway had been built through an area of comparatively low car ownership, where people were less likely than those living in other areas to benefit directly from using it. However, another resident expressed a contrasting narrative to the effect that the new motorway had helped to increase investment in the Gorbals because the area was now more accessible. Two attendees described motorway-related respiratory problems in children or grandchildren, with one child’s asthma having apparently remitted after they moved further away from the motorway. However, others suggested that the effect of the new motorway in reducing local congestion may have reduced exposure to fumes, because cars were now moving more quickly through the area rather than idling.

Govanhill event

This event took place outdoors on a main shopping street in Govanhill, which is frequented by local people and has a steady stream of foot traffic. It was not possible to set up a table or a map as at the previous events, so the stand consisted of only the study banner, and the research team simply spoke with passers-by. In total, there were seven visits to the stand over several hours.

As at the Rutherglen event, the conversation was dominated by discussion of the effects of the motorway on traffic and congestion. Those who had a car and used the motorway felt that it had reduced journey times and improved accessibility. However, some described congestion on local roads because of reconfigurations of local streets to accommodate a new motorway junction. Those who did not use the motorway expressed indifference overall, but some had concerns about the introduction of paid parking on the shopping street, which they felt had precipitated a decline in the quality of shops. Several attendees expressed the opinion that Govanhill was an area that was steadily going downhill, as a result of the in-migration of poorer, less skilled migrants and the out-migration of wealthier and more highly educated people. These changes were of greater importance to residents than any changes attributable to the motorway.

Travel and physical activity

Most attendees thought that the motorway had benefited car users. However, some felt that the improved opportunity to connect to more remote destinations came at the expense of local destinations. In addition, the new motorway was perceived to have reduced the attractiveness of other modes of transport for longer journeys, such as using the train to get to Edinburgh. Several attendees described traffic problems at motorway junctions following changes in local street configurations and traffic management problems where slip roads met local streets. The result had been increased traffic congestion on local streets and worse conditions for cyclists and pedestrians. Other, more general, traffic issues raised were those of declining public transport provision in the Gorbals, the loss of free car parking on the main shopping street in Govanhill and the generally poor quality of cycling infrastructure, all of which were issues raised at the previous community events.

Well-being

Among the local residents attending the event were a family who lived close to the new motorway. Their accounts of air and noise pollution were strikingly similar to those elicited at the first event at the sheltered housing complex. During motorway construction, piling caused structural damage to their flat and cracked nearby pavements. Even after construction was completed, the family described constant traffic noise and vibration during the day, and noise from maintenance activities at night, which disrupted their sleep. They also described breathing difficulties, which they related to increased air pollution and being unable to sit outside on their balcony.

Communities and regeneration

Some attendees felt that the motorway had created a clear visual and psychological barrier between areas, making some areas inaccessible and disrupting local social networks. Conversely, one attendee felt that the motorway had opened up views of the cityscape for those driving on it (a form of ‘reverse severance’), echoing a perspective mentioned in Chapter 7, Changes in local experience. Attendees also raised wider issues relating to social and environmental justice and inequalities. The benefits experienced by some were not thought to justify the considerable expense of the new motorway, and people of lower socioeconomic status and those with impaired mobility were seen as experiencing more harms than benefits. The new motorway was also seen as bolstering the wider economy (e.g. improving access for multinational corporations) at the expense of the local economy and small businesses. There was also more general discussion about overall decline or regeneration in local communities. Govanhill and the Gorbals were both seen as becoming ‘ghettoised’ by a combination of poorer, less skilled migrants and absentee landlords, with new estates gradually becoming slums and the social fabric of communities becoming degraded. However, this was seen as a problem that predated the new motorway. It was also suggested that although the Gorbals was benefiting from active investment and regeneration, Govanhill seemed to be languishing – a view that had also been expressed at previous events.

Road traffic accidents

Some attendees expressed the opinion that traffic had noticeably increased in areas around the M74 extension, and that it was therefore surprising that this had not been translated into more accidents. However, other attendees felt that there had been little impact on local traffic and were therefore unsurprised at the lack of change in the frequency of accidents. It was agreed that the effect of the motorway on accidents might change if congestion were to increase in future.

Changes in travel and activity patterns

After adjustment for multiple individual and household confounders as well as baseline travel behaviour, participants in our cohort survey living in the South study area surrounding the new motorway were about twice as likely as those in the North (where there was no motorway) to report travel by any mode of transport at follow-up. Within the South study area, participants living closer to a motorway junction were more likely to report travel by any mode (in the cohort analysis) and to report using a car (in the repeat cross-sectional analysis) at follow-up than those living further away. The finding on car use, in particular, was consistent with views elicited in community engagement to the effect that the new motorway had considerably improved the speed and convenience of car journeys.

The SHS data showed that the proportion of journey stages that were walked or cycled increased slightly over time in all three study areas. 63 This is an encouraging finding for public health in its own right, but there was no evidence that the rate of change was significantly different between the intervention and control areas. In contrast, our core survey data showed that the average daily quantity of active travel reported by participants decreased over time in all three areas, although this may have been an artefact of measurement as much as a reflection of a real decline. Neither area- nor individual-level exposure to the intervention was associated with either the likelihood or the quantity of active travel at follow-up in the core survey, corroborating the SHS findings.

In analogous analyses, cohort participants living in the East study area (surrounding the existing M8 motorway) were about half as likely as those in the North to report participating in any physical activity at follow-up, and those living closer to a motorway junction in the East were more likely to report a reduction in physical activity participation than those living further away. However, analysis of georeferenced accelerometer data collected in a subsample of survey participants in all three study areas at follow-up found that neither area- nor individual-level exposure to the intervention was associated with physical activity, either overall or within neighbourhood pedestrian network buffers of various sizes.

Qualitative interviews elicited a variety of personal and social factors related to physical activity, consistent with previous research. More particularly, participants described how the physical structure of the new motorway itself, as well as related changes in traffic, noise and other aspects of their surroundings, affected active travel in both positive and negative ways. Where the motorway was seen as having a direct impact, this was most often described as affecting the quality or experience of active travel rather than the frequency, duration or routes of journeys. In contrast, perceptions of personal safety were more often linked with making changes in the frequency, route or mode of active journeys. Although the divergence of views about the motorway previously elicited at baseline was reinforced, most participants at follow-up described the new pedestrian infrastructure for crossing the motorway as lighter, more open and more secure than expected.

Local residents experienced changes to connectivity that they associated with the new motorway and other factors. Broadly, those with dispersed social networks viewed the motorway as facilitating connections with amenities and people in other places. However, those whose networks were more localised, and who lacked either access to a car or confidence in using the motorway, sometimes found it be a cause of severance: physically, psychologically or through its impact on local sites of connection. More than the motorway infrastructure itself, social factors including community cohesion and perceptions of personal safety were viewed as primary causes of local severance.

Changes in road traffic accidents

The annual incidence of road traffic accidents in Glasgow City and surrounding local authorities fell by about half between 1997 and 2014. A similar reduction was observed in all three local study areas, and our analyses found no significant differences in temporal trends between study areas that could be attributed to either the construction or the opening of the new motorway. The reduction was greatest in the North study area, a pattern that appears unlikely to be attributable to the new motorway and that may reflect a variety of other factors outside the scope of this particular study. There was a social gradient in the location of accidents, and the opening of the new motorway had little impact on the spatial clustering of accidents.

Qualitative accounts of travel in the local area revealed several mechanisms that may have prevented the realisation of the claimed reduction in accidents. Local residents described new hazards, including the merging of lanes of traffic, vehicles travelling at high speeds on slip roads and altered layouts of existing streets close to the new motorway junctions. These may have contributed to the danger of crossing roads, particularly for pedestrians with impaired mobility.

Changes in well-being

Participants in our cohort survey living closer to the new motorway experienced significantly reduced mental well-being over time compared with those living further away, and in both a repeat cross-sectional analysis and a cross-sectional analysis we found evidence of an interaction whereby participants with a chronic condition living closer to the existing M8 motorway experienced a greater decrement in well-being than those living further away. In these analyses, proximity was modelled as the negative natural logarithm of the straight-line distance from home to the nearest motorway. Therefore, the average reduction in mental well-being per unit of proximity observed in these analyses – approximately 3.5 units on the MCS-8 scale – can be interpreted as the average difference between a participant living approximately 100 m from a motorway and one living 300 m away; or between a participant living 300 m away and one living 800 m away. These point estimates for the effect size on mental well-being are consistent with effects of ‘population significance’. They are similar in magnitude to those found in a clinical population experiencing reduced overall quality of life in a previous validation study (3.3 units). 66 In a general population, they are comparable to the difference between 18- to 24-year-olds and 45- to 49-year-olds (2.6 units) or between people not completing high school and graduates of tertiary education (4.2 units), and greater than that between those with a chronic physical condition and those without (2.0 units). 66 In a cohort analysis, reductions in both mental and physical well-being were also associated with an ‘upheaval index’ representing the proportion of land use in the immediate neighbourhood that had changed, particularly in the South study area. A difference of 10 percentage points in this upheaval index was associated with an estimated 1.6-unit difference in MCS-8: a difference of similar magnitude to that between people with a chronic physical condition and those without (2.0 units).

At follow-up, perceptions of both the social (collective efficacy) and physical environment of neighbourhoods were slightly more negative among those living closer to a motorway, but these differences were mostly not significant; and in the cohort analysis, there was no statistical evidence that changes in these perceptions mediated the changes in well-being associated with motorway exposure. Qualitative accounts elucidated the industrial history and character of the area, longstanding concerns about pollution and – for some – narratives of decline and powerlessness in local communities, suggesting a degree of acceptance of – or resignation to – the more recent changes brought about by the motorway. Nevertheless, the community engagement events elicited striking accounts of stark visual, noise, dust and other forms of disturbance, and their impacts on sleep, respiratory health and other contributors to well-being, although these were limited to those living in extreme proximity to the new motorway (see Explaining the effects of the intervention).

Estimating the effects of the intervention

Explaining the effects of the intervention

Although estimating effect sizes forms an important part of causal inference in natural experimental research, it is also – and, some would argue, equally – important to investigate potential explanations linking interventions with their observed effects, or lack of effects. 89 In this section we draw on insights from our qualitative research in particular, and from our qualitative and quantitative findings in combination, with two main aims. The first is to strengthen – where appropriate – the causal interpretation of the main outcome analyses discussed above, which can never be entirely free from the threat of residual confounding. The second is to investigate the extent to which the findings of this study may be generalisable.

Testing more generalisable causal hypotheses

We now turn to the potential contribution of our findings on causal estimation and causal explanation to more generalisable causal inference. It is an inescapable reality of this type of place-based natural experimental study that we cannot assume that effects we have observed in selected neighbourhoods in Glasgow would necessarily be replicated elsewhere in Scotland, the UK or further afield. Indeed, no single intervention study in this, or any other, field of applied public health research is likely to be capable of supporting ‘generalisability’ in that narrow sense. On the contrary, the only way to generate such inferences is by combining evidence from multiple studies. To that end, our findings begin to corroborate and extend those of other studies in showing the ways in which the behaviour, well-being and health of populations may respond to changes in the built environment, and will in time be accumulated with those of other current and future studies in a programme of evidence synthesis that lies beyond the scope of this report. In this section, we therefore focus on drawing together the various strands of this research project to assess the degree of causal support they offer for the overarching hypotheses that motivated the study and were outlined in Chapter 1 of the report. 87

We articulated the potential effects of the new motorway using two contrasting overarching vignettes of a ‘virtuous’ and a ‘vicious’ spiral (see Table 1). Mapping our findings against the key propositions of each vignette, we find – perhaps unsurprisingly – a mixture of confirmatory and disconfirmatory evidence on both sides. In respect of the ‘virtuous spiral’ hypothesis, other published data suggest that traffic on local roads has indeed been reduced, but we found only partial qualitative evidence that this had made conditions more pleasant for pedestrians and cyclists (at certain locations), and no evidence that people perceived their local environment in a more positive light or spent more time out and about in it as a result. We found divergent evidence about perceptions of noise and air pollution, with some people clearly experiencing marked adverse effects in this regard and others – as predicted – not noticing these changes against the background conditions of their urban surroundings. We found no evidence for an increase in either well-being or physical activity and comparatively strong evidence for a harmful effect on mental well-being. In respect of the ‘vicious spiral’ hypothesis, we found qualitative evidence that traffic had increased on some local streets and comparatively strong evidence that local people were more likely to travel – and by car – than they would have been in the absence of the new motorway. This and other qualitative evidence also provides some preliminary support for the predicted ‘car-bound exodus’ from ‘declining’ local areas. We found some support for the predicted degradation of the local environment making conditions less pleasant or safe for pedestrians and cyclists, but only for some people and not in our quantitative analyses.

Given the mixture of confirmatory and disconfirmatory evidence for either the ‘virtuous’ or the ‘vicious’ spiral, we went on to address the challenge of synthesising causal inferences from across this body of findings by taking a ‘pragmatic pluralist’ approach to the ‘ragged evidence’ of the natural experimental study. In so doing, we sought to match patterns of outcomes with patterns predicted by the intervention theory imperfectly captured in these vignettes, searching not for support for a singular overarching hypothesis, but rather for the least implausible explanation of the conditions that may be required to produce or prevent the outcomes of interest. 53,87,132,151

Turning first to the ‘virtuous’ spiral, although we did find some (mostly qualitative) evidence for beneficial impacts, our analysis suggests that the realisation of these benefits may depend on a mixture of individual, social and environmental factors in the causal recipe. The qualitative data suggest that more neutral or disinterested responses may be more likely where new infrastructure, and perceived associated pollution, is consistent with existing local narratives of place (as post-industrial or scenic and natural; as polluted or clean; or as a characterised by change or stability) or in areas where residents perceive themselves as having low collective efficacy. 152,153 In addition, local impacts of new transport infrastructure may be superseded by other more immediately pressing concerns, for example in areas of high perceived crime or significant change in social composition. Beneficial outcomes may be more likely where new urban transport infrastructure ameliorates, or at least does not exacerbate, the severance of residents from people and places nearby; where complementary actions are taken to ‘lock in’ early benefits, for example by calming or diverting traffic; where the intervention enhances the connectedness of people’s social life worlds, rather than disrupting them; and where associated pedestrian infrastructure minimises pre-existing negative social factors such as fear of crime. For those with more dispersed social networks and the means to travel by car, this last mechanism may be evoked by improving the major road network, and the journeys those people make further afield could contribute to their physical activity, prosperity and well-being. However, there are other social and environmental arguments in favour of encouraging more localised activity patterns, and the more ‘hypermobile’ society that is facilitated by policies that favour motor vehicle use may have very considerable opportunity costs. 154,155

Those opportunity costs are exemplified in the ‘vicious’ spiral, which can now be described in terms of a plausible causal recipe for adverse, inequitable impacts. Harmful outcomes may be more likely where the infrastructure further divides local communities: not only in the tangible sense of introducing physical obstacles to local journeys, but also in the way it partitions local people into those who gain and those who lose. Where people perceive their community to be in decline, or otherwise changing in ways that make them feel uncomfortable, and where they lack the means to benefit from the improved transport network themselves – or the means or autonomy to relocate to a more pleasant environment – this type of urban redevelopment may further erode a declining stock of social capital and sense of efficacy among more disadvantaged local residents and increase the relative burden of ill-health in those people. Even if investment in new major road infrastructure can be shown to bring wider economic benefits and is justified – as are many public health interventions – on utilitarian grounds, the question remains whether or not alternative ways of investing the same money might have produced more equitable and sustainable benefits for local people.

Ultimately, rather than conceptualising the new motorway as a complex set of inter-related and highly specific changes to the built environment – new carriageways, viaducts, slip roads, pedestrian crossings and the like – it may be more meaningful, both for understanding causal mechanisms and for drawing lessons for policy and practice, to shift analytical attention away from the surface form of the intervention towards its underlying function. 156 In this case, our evidence points to two critical functions – connecting and separating – that constitute two sides of the same coin and are both evoked by the same intervention in different ways for different people. Following a pragmatic pluralist, pattern-matching approach rather than a binary hypothesis-testing approach to causal inference, the overarching hypothesis with which our data are most consistent is that new transport infrastructure is more likely to benefit more people when it connects people with their social and physical surroundings – broadly defined – more than it separates them, and when people are protected from its harmful environmental impact by distance or other effective mitigation measures.

Interpretation for public health policy and practice

Living near to a new motorway was associated with a substantial worsening of local residents’ mental well-being, particularly for those exposed to a greater degree of upheaval in their immediate surroundings. Sources of pollution such as major roads and industrial sites are already disproportionately located in or near to more disadvantaged neighbourhoods, and our findings indicate a clear potential for such disadvantage to be amplified by further disrupting such neighbourhoods with new major road infrastructure. The additional concentration of this disadvantage among those with a chronic condition living near an existing motorway shows how one form of disadvantage (environmental) may be compounded over time by another (health), thereby contributing to worse clinical outcomes among those whose health is already impaired and who lack the means or autonomy to move house. This type of inequitable and harmful health impact, and its implications for the costs of health and social care, should be considered in the appraisal of proposals for future infrastructural investment, particularly in areas characterised by high levels of social housing, material deprivation and ill health.

Although we found little clear quantitative evidence for effects on active travel, physical activity or sedentary behaviour 2 years after the motorway was opened, that is not to say that the effects of building new major road infrastructure or its implications for policy and practice are neutral in this respect. The new motorway appeared to promote travel in general and car use in particular among those living nearby. Particularly in a country such as Scotland in which car ownership and use are strongly socioeconomically patterned, this direction of effect is liable to reinforce existing inequalities in access to amenities and opportunities, and runs counter to the stated objectives of national transport and health policies in Scotland, England and numerous other countries. The appraisal of proposals for future infrastructural investment should consider a wider range of options than the two considered in the PLI for the M74 extension, namely either to build it or not to build it. A more wide-ranging option appraisal might have explored the comparative environmental, health, social and transport benefits and costs of alternative uses of the resources, and this might have led to a different decision. Similar considerations apply, with perhaps greater urgency, in (mostly lower- and middle-income) countries and regions currently going through a ‘motorisation transition’ involving the rapid construction or expansion of highway networks.

Our findings complement those of a previous systematic review in suggesting that policy-makers cannot necessarily appeal to a probable reduction in the incidence of road traffic accidents to justify the construction of new major roads. 51 It may be taken for granted that such projects alone will reduce casualties, as was argued prior to the construction of the M74 extension, but we found no evidence that any such benefit had been achieved by 2014.

In summary, there is currently little clear public health evidence to guide decisions about investing in expensive urban regeneration projects of this kind. In terms of well-being, motorway construction appeared to be more harmful than helpful to local communities. We acknowledge that more time may be required for some benefits, such as economic revival (which we have not assessed directly), to be fully realised and to produce a more indirect positive impact on well-being. Nevertheless, it cannot be assumed that the benefits of regeneration will necessarily outweigh the harms of the physical and social impacts of the new infrastructure on local communities. In an area with an existing motorway, those with chronic conditions experienced the greatest adverse effect on well-being. This implies that transport projects of this kind may entrench existing health inequalities in an already vulnerable population. From a social justice perspective, there did not appear to be a fair distribution of benefits afforded by the motorway compared with its associated burdens among those living nearby, particularly for those without access to a car. Although the specific impact of environmental changes of this kind might very well be experienced differently in communities lacking an industrial history or an existing plane of severance, our findings clearly indicate a more general implication for policy and practice, namely that some of the overall health and social gains claimed for this type of investment either may not be achieved or may be achieved for some at the expense of others.

Challenges of the study

We faced a number of challenges in the design and execution of this study. We have chosen to highlight four of these here because they illustrate some of the difficulties inherent in natural experimental studies in public health intervention research. Although some of these issues were raised in a study design paper published in 2006,52 shortly after the baseline study began, they have become more widely discussed in the public health literature in the intervening decade, particularly in – and following – the guidance published by the MRC in 2011. 59

The first challenge was that of designing a tractable and independent public health study of an apparently simple, but in practice complex and highly contextual (and politicised), intervention in the urban environment. The public discourse about the proposal to extend the M74 motorway elicited many possible outcomes and pathways to those outcomes, some of which might be expected to take many years to develop and all of which might be worthy of evaluation in their own right. We acknowledge that we could have set out to evaluate the impact of the project in a more all-encompassing way, but instead we chose to focus our evaluative resources on a more limited – and more tractable – set of research questions that we could address within the context of a larger research programme and portfolio of related studies. We worked hard to maintain scientific independence and objectivity throughout the study, from the delineation of the main competing hypotheses onwards. When the study was originally conceived, the principal investigator lived not far from the boundary of one of the study areas, and the familiarity of several members of the research team with the local context was undoubtedly helpful in formulating the study design, achieving rapport with study participants and interpreting their accounts. However, any public health researcher living in the vicinity would naturally be expected to have their own attitudes and beliefs about the controversial topic of the study. We therefore ensured that survey materials at both baseline and follow-up were presented in terms of a general survey on ‘traffic and health’ without any reference to the new motorway, that the follow-up qualitative fieldwork was conducted by a researcher with no local connection; that the topic of the new motorway was allowed to arise naturally in the course of interviews, rather than being presented as their main purpose; that participants were encouraged to discuss both positive and negative impacts of the new motorway; and that minority opinions were identified and reported.

The remaining challenges were more methodological or technical in nature. The second challenge was that of defining exposure, both to the new motorway itself – the intervention of central interest to the study – and to related changes to the environment. Although simple area-based exposure measures are commonly used in natural experimental research, in this study our decision to augment these with individual measures a priori was vindicated by the finding that the latter measures proved to be more convincingly related to some of the outcomes. 157 Graded exposure measures of this kind have also been used in other recent natural experimental studies in this field, but the optimal definition of exposure for a given study is likely to vary according to the particular combination of intervention, study design and outcome in question and, therefore, to require careful thought in all cases. 80 The third challenge was that of delineating a comparison area or areas, in order to provide some evidence of what might have been expected to happen under a counterfactual scenario of no substantial change to the environment. Comparatively well-balanced study areas and samples were achieved at baseline as a result of painstaking work, but it is simply unrealistic to imagine that all important attributes of an area and its residents – including its road network, its topography and its history – could be precisely ‘matched’ in any number of control areas. Rather than seeking corroboration between comparisons of outcomes with multiple external control areas, we chose to use individual measures of exposure within our study areas as an additional basis for controlled comparisons, and to rely as much on causal explanation as on causal estimation as a way of investigating the evidence for causal inference. However, this corroborative approach threw up a fourth challenge: that of reconciling differences in outcomes between analyses using different samples and exposure measures. We did not expect perfect agreement in this respect, but our experience highlights the importance of thinking carefully about the different ways in which interventions may lead to outcomes in individuals and in populations, and about the complementary strengths, limitations and meanings of different analyses in this respect.

In the following sections, we briefly discuss our responses to these four challenges in terms of their implications for the strengths and limitations of the study and for future research, most which have been described elsewhere in the report but which are summarised here for the sake of completeness.

Strengths of the study

This is one of few intervention studies to have examined how changes in the environment influence changes in health and well-being, particularly with a focus on active living in a comparatively deprived population. It thereby adds to a small but growing body of evidence for the effects of changing the built environment on patterns of travel behaviour, physical activity, accidents and well-being, as called for in recent reviews arising from the policy, practice and research communities alike. 38,44,158 More specifically, it is the first study to have evaluated the effect of a new motorway on physical activity and sedentary behaviour, and it makes a particular contribution to our understanding of the impacts of new major road infrastructure by investigating changes in multiple related outcomes at both individual and population level.

We used two extensively validated instruments to capture changes and differences in various aspects of well-being, and assessed overall physical activity using the short IPAQ, which, although admittedly not ideal for detecting change in the time spent in different activities, has been widely used around the world and has comparable criterion validity to that of any other physical activity questionnaire of similar length. Although our measures of travel behaviour were not formally validated, our own survey instrument was closely based on that used in the National Travel Survey,64 and in both our own data and the SHS data63 we were able to disaggregate travel to the level of the mode of transport used for each stage of a journey. This enabled us to ascertain, for example, walking or cycling as part of a longer journey. Among other things, our study illustrates the potential value of using routinely collected data in the evaluation of natural experiments, most obviously in respect of the police STATS19 accident data,74 but also in respect of the SHS travel diary data set that we analysed. 63 This included > 20,000 journey stages reported by nearly 8000 people in the west of Scotland, collected as part of an ongoing repeat cross-sectional survey of a large, random, representative national sample of households, which we were able to analyse at minimal marginal cost.

Ours is among the first studies to have evaluated the effects of any environmental change on this range of outcomes using robust quasi-experimental methods, as outlined in research recommendations from NICE. 37 We collected original survey data from 2323 local residents, supplemented with detailed activity measurement in a subsample of 196 and qualitative fieldwork in a subsample of 30 (as well as 12 other key informants), along with analysis of 78,919 road traffic accidents and 7911 travel diaries from other routinely collected data sets. We investigated how the outcomes were associated with multiple measures of exposure, objectively defined at both area and individual level using a GIS. We made comparisons both between and within areas, in both longitudinal and repeat cross-sectional analyses, combining these different quantitative analytical approaches with each other and with qualitative research methods to offset the limitations of each approach and corroborate findings where appropriate. We controlled for a set of potential confounders in all models. A further strength of our modelling strategy lies in the use of two-part models for travel and activity outcomes that were not normally distributed. Despite their good theoretical and statistical fit and efficiency in combining analytical options, such models have seldom been used in this field to date. The ARIMA procedure used in our time series analyses provided a rigorous way of assessing impacts on accidents that are measured using count data. Like all statistical models, these analyses depended on assumptions about the form of the relationship between the intervention and its effects, but sensitivity analyses – in which key parameters of the time-series and other statistical models used in the study were altered – suggested that the main findings were not sensitive to these assumptions. Finally, the study provides a rich data set that constitutes a valuable platform for further analyses, about which potential collaborators can find out more at our data sharing portal. 159

Limitations of the study

It was not feasible to collect data relating to the main outcomes of the study throughout the period of motorway construction over a longer follow-up period, or with more detailed consideration of people who had moved into and out of the study areas. It is possible that – with additional funding – extending data collection in any of these ways might have revealed additional short-term impacts, the maintenance (or otherwise) of effects observed at 2-year follow-up or the emergence of other impacts in the longer term. Neither was it feasible to study the full range of potential impacts identified in public discourse, either before or after the motorway was constructed. In particular, we acknowledge the importance attributed to noise, air pollution and wider economic impacts by participants in our community engagement activities, which we were not able to assess directly in our own research. The evidence of impacts that we have elicited despite these limitations indicates the value of seeking to replicate our longitudinal analyses in larger samples with a longer duration of follow-up where possible, particularly in situations in which outcomes might be ascertained using routinely collected surveillance and population survey data.

We extracted data on road traffic accidents from the STATS19 portal74 and, although these data are widely used in research, they have known limitations. Chief among these are that they rely on accidents being reported to the police, and on the accuracy of police officers in reporting their exact locations. However, systematic differences or changes in the accuracy of reporting between study areas appear unlikely, particularly given that a single police force covers the region. Other limitations of our data include the collection of only 1 day of travel data for each participant at each time point, which provide estimates of activity with greater variance than those derived from longer (e.g. weekly) periods of measurement, the reliance on simple self-reported measures for estimating changes in travel, physical activity and sedentary behaviour over time and the reliance of the mediation analyses on self-reported measures of both the social and physical environment of neighbourhoods that may not have captured the mechanisms of most importance in mediating the outcomes of interest. The analysis of SHS travel data was subject to two additional limitations. First, journey distance in that data set is based on the straight-line distance between two points, rather than the inevitably longer real distance along the road and path network. Second, in the interests of data protection, participants in those analyses were assigned to study areas according to their area of residence, rather than to the more precise location given by their full unit postcode.

We chose to include a repeat cross-sectional design because we anticipated considerable attrition of the cohort over time, and this was confirmed, with approximately 70% of the original baseline sample lost in the 8 years that elapsed before follow-up. Together with the comparatively low response rate of approximately 16% at each time point, this represents an important limitation on the extent to which the survey data could be considered representative of the local population. The response and attrition rates were comparable to those of similar natural experimental studies in similar settings,6,49,160 and the repeat cross-sectional design was chosen to offset the anticipated attrition in particular. Nevertheless, the combination of large variances in some of the outcome measures, the comparatively small cohort sample, and the comparatively small number of pedestrian and cyclist casualties all contributed to limiting the statistical power of the study to detect some effects, although it is equally plausible that the lack of significant findings for those outcomes may reflect a true absence of associations. The scope of the qualitative study of residents was also limited by the achieved sample size of 30 participants. Although this does not necessarily threaten the validity of the inferences, it is possible that further themes may have been elicited and explored if more people had taken part.

A final limitation is the possibility of residual confounding, either because of an unmeasured confounder that may have been unequally distributed in the study population at baseline, or because of other regeneration projects or concurrent changes in the built environment, such as the wider Clyde Gateway initiative and the 2014 Commonwealth Games. This is a core and unavoidable challenge of this type of natural experimental research, and is complicated by the fact that the effects of a given ‘intervention’ may depend on its being related to those other changes rather than being observed merely as a result of ‘contamination’ by them. We sought to minimise the impact of potential residual confounding by carefully delineating the study areas to ensure the comparability of the samples and settings, by adjusting the analyses for multiple sociodemographic covariates, and by comparing the results of complementary analyses with different strengths and limitations. We did find some sociodemographic differences between the follow-up samples in the three study areas, despite the considerable care taken to balance the samples at baseline. Even with randomisation, perfect balance is not always achieved, and in this case it certainly remains possible that other actions taken in some neighbourhoods may have directly or indirectly influenced local patterns of activity and well-being. To take one example, we observed an unanticipated finding of an increased likelihood of bus use over time in cohort participants living in the East study area. It seemed unlikely that this outcome, observed between 2005 and 2013, could be causally attributed to a motorway built in the 1960s and more likely that it might reflect concurrent improvements to local bus services. An initial scoping exercise indicated that public transport improvements had taken place in the area, but the scale of work required to comprehensively assess and quantify this additional time-varying exposure was beyond the scope of the current study. Nevertheless, our study does represent a methodological advance on the uncontrolled or poorly controlled studies typical of previous research in this field, and other aspects of its conceptualisation and design provide a degree of complementary support for the causal inferences made.

Methodological investigation and development

Although methodological research was not the primary aim of this study, its design, evolution and execution over more than a decade have provided considerable methodological learning that may be of more general value to the field. We have shown how exposure to a particular physical environmental change may be characterised and analysed using a combination of area- and individual-level measures. We have also devised a novel measure of more general physical environmental change, which we have described as an upheaval index, and we have developed and demonstrated a method for computing this using aerial imagery in the public domain. We have applied both quantitative and qualitative research methods that are comparatively little used in this field, notably two-part regression models and the spatial method of photovoice interviews, adapting these to the particular needs of a natural experimental evaluation and illustrating ways in which they might be useful in future studies. We have sought to respond to a number of insights from critical scholarship as it relates to evaluation research, as exemplified by our use of vignettes to capture overarching research hypotheses linking multiple outcomes and mechanisms, and our use of the notion of pattern matching as a way of synthesising evidence for and against competing hypotheses. 53,87 We do not claim to have done more than scratch the surface of the problem of making sense of complex evaluation data in this way, but we hope that our efforts will contribute to a more general movement towards more thoughtful and ultimately generalisable public health intervention research. Finally, we have tried to embrace the perspectives of local communities in the research design in a meaningful way, partly in helping to shape our qualitative fieldwork but also – and perhaps more significantly – in contributing qualitative causal process observations to our understanding of intervention mechanisms (see Explaining the effects of the intervention), ‘ground-truthing’ our emerging findings and seeking to produce a credible interpretation of those findings, thereby feeding into a wider discussion about how research may be used by and be useful for communities.

Under review

Olsen J, Foley L, Ogilvie D, Mitchell R, Mutrie N, on behalf of the M74 study team. Population levels of, and inequalities in, active travel: a national, retrospective, cross-sectional study of adults in Scotland. Under review.

Prins R, Foley L, Mutrie N, Ogilvie D, on behalf of the M74 study team. Effects of urban motorways on physical activity and sedentary behaviour in local residents: a natural experimental study. Under review.

In preparation

Foley L, Coombes E, Humphreys D, Jones A, Mitchell R, Ogilvie D, on behalf of the M74 study team. Effects of exposure to change in the neighbourhood built environment on the wellbeing of local residents in deprived areas: natural experimental study. In preparation.

Nimegeer A, Hilton S, Thomson H, Foley L, Ogilvie D, on behalf of the M74 study team. Perceptions of the effects of a new urban motorway on active travel: qualitative evidence from a natural experimental study. In preparation.

Nimegeer A, Hilton S, Thomson H, Foley L, Ogilvie D, on behalf of the M74 study team. Severance and connectivity in the wake of a new motorway: qualitative evidence from a natural experimental study. In preparation.

Olsen J, Mitchell R, Ogilvie D, on behalf of the M74 study team. Spatial-temporal trends in road traffic accident clusters and the impact of new transport infrastructure; a retrospective longitudinal natural experiment, Scotland, UK. In preparation.

Experiences and perceptions of neighbourhood

• The local area as a place to live (general).

• The local area as a healthy space.

• Access to amenities.

• Suitability for children and families.

• Suitability for teenagers.

• Suitability for older people.

• Feelings of safety (general).

• Feelings of safety and noise (traffic related).

• Feelings of connectedness to other neighbourhoods.

Changes in experience and perceptions of neighbourhood

• Has respondent’s experience of living in the local area changed? If so, how?

• Have respondent’s perceptions of or feelings towards the local area changed? If so, how?

• Why does respondent feel that these changes have come about?

• Have changes in the physical environment influenced how the respondent experiences and perceives the local area? If so, which specific physical changes?

Use, travel and activity within the local area

• Types of activity undertaken in local area (social, work, visiting friends, shopping, exercise, etc.).

• How often these activities are undertaken.

• Types of travel within local area (how do participants get around?).

• Does this include active travel (e.g. travelling to shops, work, to visit friends)?

• How often does active travel occur?

Changes in use of the local area

• Have types of activity undertaken in local area changed recently? If so, how?

• Has active travel changed, increased or decreased? If so, how?

• Do respondents feel any more or less comfortable travelling on foot or by cycle in local area than they used to? If so, describe.

• Explanation for changes: why do they think changes have occurred?

• Have changes in the physical environment influenced how respondent uses the local area? If so, which specific physical changes?

If it has not been mentioned yet at this stage, explicitly ask about the impact of the M74 extension.