Feasibility of assessing public health impacts of air pollution reduction programs on a local scale: New Haven case study.

Assessing the overall impact of cumulative air pollution programs on environmental public health is a daunting task. Air quality has improved substantially in the United States in recent decades, in large part because of increasingly stringent federal and state air quality regulations. Although many studies have documented links between better air quality and improvements in a variety of human health metrics (Clancy et al. 2002; Friedman et al. 2001; Hedley et al. 2002; Heinrich et al. 2002; Pope 1989; Pope et al. 2009), direct evidence concerning the extent to which specific control measures have improved health is lacking. This lack of evidence is due in part to inherent difficulties in environmental health research concerning effects of relatively low-level exposures on multifactorial health outcomes with long latency periods that often are associated with small changes in relative risk (RR). In addition, complex interactions between interventions over time make it difficult to isolate the environmental health effects of any one regulation. No single study or study design is likely to characterize the entire range of public health improvements attributable to an air quality regulation (Health Effects Institute 2003). Although the study of the impact of cumulative air pollution control programs on environmental public health is fairly unique compared with those involving individual programs, and formal research approaches are still in early phases of development, several recently published research studies have contributed relevant insights by taking advantage of episodic, one-time events or natural experiments such as a coal ban (Clancy et al. 2002), traffic reductions (Friedman et al. 2001; Tonne et al. 2008), and closure of a steel mill (Parker et al. 2008; Pope 1989). Burnett et al. (2005) proposed a “measure of progress” equivalent to the percent reduction in the RR of adverse health outcomes attributable to reductions in ambient air pollutant concentrations. They note that their proposed approach is applicable to single and multiple cities, as well as single and multiple pollutants. However, they demonstrate their approach only with time-series data on mortality associated with ambient nitrogen dioxide (NO2) concentrations from 1981 to 1999 in multiple Canadian cities. Jerrett et al. (2007) pointed out that evidence of short-term health improvements based on studies of “natural experiments” and well-delineated interventions may not apply directly to effects of gradual air quality improvements over many years, and when the follow-up period is long, many other factors, such as random error and systematic biases, especially for low RRs (< 1.05), can obfuscate the linkage of air quality improvements to health benefits. Thus, it is important to determine a priori which pollutant–outcome relationships are most likely to result in observable impacts on health in a particular population given projected changes in air pollutant concentrations and estimated risks associated with exposures of concern. Of course, even when benefits are not statistically detectable based on observational data, it is helpful to decision makers to know the circumstances in which health benefits of pollution reductions are expected to occur. This challenge was recently addressed by Pope et al. (2009) for ambient particulate matter (PM). They used an ecological analytic approach whereby they examined associations between noticeable changes in ambient PM levels and differences in life expectancy estimates [~ 0.61 years per 10-μg/m3 decrease in PM ≤ 2.5 μm in aerodynamic diameter (PM2.5)] across multiple metropolitan areas in the United States between 1980 and 2000 using multiple regression models adjusted for socioeconomic status, demographic characteristics, and smoking. The models were similar in structure to those previously used in a cross-sectional analysis of air pollution effects on mortality by Ozkaynak and Thurston (1987). This approach can be applied at a much smaller geographical or urban scale if the necessary health, air pollution, demographic, and other explanatory data can be accessed or estimated at fine-scale spatial resolution. In the present study, we used both conventional sample-size–based approaches and the strategy presented by Pope et al. (2009) to assess the feasibility of conducting a study on the impact that cumulative air pollution reduction programs may have on environmental public health within a small geographic area, New Haven, Connecticut (USA). We then evaluated the strengths and limitations of these approaches in the context of urban scale assessments. We chose New Haven for this assessment because it was designated as one of two Connecticut counties in nonattainment of the PM2.5 standard in 1997 [U.S. Environmental Protection Agency (EPA) 2010], and because a variety of air pollution reduction activities have subsequently been implemented at multiple jurisdiction levels by various federal, state, local, and voluntary actions, including the CARE (Community Action for a Renewed Environment) program (U.S. EPA 2011). National and regional initiatives have resulted in large reductions in ambient nitrogen oxides (NOx) from mobile sources in New Haven, and the Northeast region also adopted more stringent vehicle emission standards earlier than did other parts of the United States and had faster fleet turnover. In addition, New Haven has implemented a number of voluntary air pollution reduction activities such as promoting smoke-free homes, use of ultra-low-sulfur diesel fuel, school bus retrofits, solvent reduction workshops, and Tools for Schools (City of New Haven 2004).