Table 4

Case study of the dry stone wall principle applied to a systematic review

Intervention consideredMain research questionKey principles of approach takenExamples of key methods and findings
‘Environmental changes aimed at encouraging walking or cycling’31‘To understand how changes to the external physical environment may act to promote walking, cycling and physical activity, and why these may or may not be effective’Looking beyond interventions
‘Rather than synthesising evidence from similar classes or forms of interventions (eg, cycle paths), it might be possible to synthesise evidence from interventions which have the same function (eg, interventions which change the perceived safety of cycling regardless of the precise method used). This exploits the variation in contexts where similar (but not exactly the same) interventions have been implemented.’‘We identified three common resources that interventions provide to promote walking and cycling: (1) improving accessibility and connectivity; (2) improving traffic and personal safety; and (3) improving the experience of walking and cycling.’
Searching for patterns
‘In the spirit of triangulating a range of types of evidence, we used principles from a range of different methods including narrative and realist reviews and qualitative analysis as recommended and used a sequential explanatory approach (…) We extracted information on the evidence for effects (‘estimation’), contexts and mechanisms (‘explanation’) and assessed credibility, and synthesised material narratively (…)’‘We found some evidence that interventions were considered with the wider physical and social system in policy documents and qualitative or mixed-method studies. These sources of evidence are traditionally viewed as lower quality, and although they were few in number here, we found that they were useful in painting a conceptually richer picture of potential contexts and mechanisms.
Embracing the mess
‘We identified common functions—overarching themes—across these interventions (and) synthesised (…) combinations of contexts, mechanisms and outcomes on a more abstract level (…) with a focus on exploring patterns of outcomes (more successful and less successful) and on those with the strongest or most convincing evidence (…) We distilled three potential ways in which the interaction of an intervention’s function with different contexts may lead to processes and outcomes being enabled or disabled.’‘The most plausible mechanisms concerned (1) improving accessibility and convenience of walking and cycling, and (2) reducing potential conflict between users (…) The most effective interventions appeared to target accessibility and safety in supportive and unsupportive individual and physical contexts.’
  • Emphases added.