Table 3 Summary of suggested changes to causal loop diagrams (CLDs) and specific modifications in the diagrams.

Although there are many indicators available, they are not always the right indicators. Participants suggested adding a negative arrow from ‘Perceived relevance and authority of UHI tools’ to ‘Production of new UHI tools’.

The negative arrow was added, creating feedback loop R9 in Fig. 2.

The process of UHI tool institutionalisation was seen as a key factor that led to culture shifts within organisations and wider knowledge, which changed conversations and policy about urban health over time.

An arrow with positive polarity was added between ‘Institutionalised UHI tools’ and a new variable ‘Culture of health-promotion’ in Fig. 5.

Participants wanted to include a variable on actual health outcomes within the model to indicate whether UHI tools improve health.

An urban health variable was re-introduced (from a previous model iteration) to Fig. 4 as ‘Urban population health’ with a positive polarity arrow from ‘Health-promoting built environment’.

An indicator user cautioned that the model did not show the potential danger of applying single indicators out of context. They said that this can lead to oversimplification of problems, which can cause unintended consequences.

Structure was added to Fig. 4 in the form of two additional variables ‘Misapplication of single indicator’ and ‘Knowledge of policy impact’, with links to existing variables from the high-level CLD ‘Wider knowledge of urban health’ and ‘Use of UHI tools’.

An indicator producer felt that the different potential uses of UHI tools, such as baseline assessment or policy-monitoring, were not represented in the model.

Early versions of the model contained variables such as ‘Monitoring of urban health’ and ‘Knowledge of policy effectiveness’ (later combined). The latter appeared in early versions of the model as an effect of ‘Use of UHI tools’ that led to ‘Targeted health-promoting policies’. Simplified versions of the model removed this variable. However, following the topic of ‘Misapplication of single indicator’ (above), the ‘Knowledge of policy effectiveness’ variable was re-introduced as ‘Knowledge of policy impact’.