Abstract
Theory and modelling studies suggest that some interventions to decarbonise emissions in the transport sector can, in addition to the long-term benefits from contributing to stabilising the global climate, have substantial short-term benefits for population health. Policies that encourage active modes of transportation, for example, may increase population physical activity or decrease air pollution, thus reducing the burden of conditions such as colon cancer, diabetes, heart disease and dementia. This thesis was concerned with examining the evidence to support these co-benefits through two pieces of research; a systematic review of evidence and a natural experiment analysis.
The systematic review searched for evidence on whether policies that decrease transport sector CO2 emissions have a measurable effect on health determinants, population health and/or health inequalities under ‘real-world’ conditions. Twenty-two studies of eleven relevant interventions from four countries were identified after the review process. These interventions were mostly of personalised travel planning and showed modest increases in active transport, such as walking, and reductions in vehicle use and CO2 emissions. The majority of the studies identified in the review were poor quality, and, with one exception, the validity of the reported results cannot be assumed.
The natural experiment analysis utilised the variations in petrol and diesel price in New Zealand to explore whether increases in price have an impact on transport-related air pollution. This was intended to be a ‘natural experiment’ or proxy, to estimate whether an effective carbon price would have a discernible impact on health outcomes through reduced air pollution. A causal diagram was used to inform the analysis. Linear regression was conducted on a weekly time-series of NOx, from four roadside air quality monitoring stations around New Zealand, and fuel pricing data, from 2001 to 2013. Splines were used to remove seasonal and long-term trends. The model was adjusted for the presence of a public holiday during the week. Exploration of the lagged price exposure and NOx outcome was also undertaken. Fixed effects meta-analysis of results from the four stations was undertaken, as were multiple sensitivity analyses using different methods of removing trends, co-variates, and air pollutant outcomes.
The adjusted models, for all except one air quality station, showed a modest reduction in NOx associated with an increase in (unlagged) petrol price. For example, data from Khyber Pass air quality station a 1% increase in petrol price was associated with a 0.71% (95% CI -3.04 to 1.61) reduction in NOx. All confidence intervals included the null. The lagged models (with each lagged petrol exposure entered into a separate model) did not show consistent patterns in all four stations however two stations had (non-significant) patterns suggestive of an initial drop in NOx after a price rise, followed by a rebound increase, resulting in little net change over the two month period studied. This pattern was also seen in the meta-analysis, although many of the estimates remained non-significant. The findings for the diesel price and NOx association were similar, but attenuated, to petrol price and NOx associations.
Sensitivity analyses showed similar associations for all other air pollutants (PM10, PM2.5, NO2, and CO). Including meteorological co-variates did not consistently alter the direction or statistical significance of the association. The association between fuel price and NOx was sensitive to the method and amount of de-trending used in the model.
The existing body of real-world evidence to support health co-benefits of decarbonising the transport sector is limited. This analysis provided some support for the existence of a short-term association between fuel price changes and transport-related air pollutants, although the net impact over the time period studied was probably neutral. There were issues with analysis power, de-trending techniques and construct validity of the results. There are methods to improve the breadth and quality of evidence in the area of decarbonisation co-benefits. However, the complexity of the area and the need to act quickly to reduce emissions means that there may be ongoing uncertainty about both the carbon and health impacts of many policies.