NiNiProblem statement
Heat-health research reliably shows that hot days raise emergency demand. City planners cannot act on that finding, because they cannot change the weather; they can change shade. The policy-relevant quantity is therefore not the temperature-mortality slope but the marginal effect of canopy cover at a fixed temperature.
Research questions
- RQ1. Does neighbourhood canopy cover modify the association between daily maximum temperature and heat-related emergency visits?
- RQ2. Is the modification larger in neighbourhoods with older residents or lower air-conditioning prevalence?
- RQ3. What canopy increase corresponds to a clinically meaningful reduction in visits on a 35 °C day?
Design
A time-stratified case-crossover design over five summers, in which each patient serves as their own control on nearby non-event days. This removes all time-invariant individual confounding — age, comorbidity, housing — without requiring it to be measured.
Exposure combines station temperature, satellite-derived land surface temperature at 30 m, and canopy cover from municipal LiDAR, assigned by residential geocode.
Analysis plan
Conditional logistic regression with distributed lag non-linear terms for temperature over zero to three days, interacted with canopy quartile. Effect modification by age band and by an area-level air-conditioning proxy is pre-specified rather than explored.
Limitations
Residential geocodes misassign exposure for people who work elsewhere during the day. Canopy is measured once and treated as static within a summer. Emergency visits undercount heat morbidity managed in primary care.