Reductions in Respiratory Hospital Visits after a Coal Coking Plant Closure: A Natural Experiment

Abstract

Rationale: Abrupt air quality improvements have followed the closure or dramatic emission control of large air pollution sources. These "natural experiments" provide ideal opportunities to assess the real-world health benefits of air quality improvements. The shutdown of the Shenango coking plant, a significant fossil-fuel pollution source located on an island in the Ohio River near Pittsburgh, Pennsylvania, presented such an opportunity to test for changes in respiratory health in the local community after the closure. Objectives: We sought to identify and quantify the immediate and/or longer term changes in respiratory hospitalizations and emergency department (ED) visits among the population residing near the Shenango coke plant at the time of its closure. Methods: We acquired data for respiratory hospitalizations and ED visit counts from residents living in zip codes surrounding the plant, as well as at comparison control sites, 3 years before and after the shutdown date. The immediate and longer term changes of respiratory health outcomes were tested with an interrupted time series model and compared with findings from external control sites and internal control outcomes. Measurements and Main Results: We found that the closure of the Shenango plant was associated with an immediate 20.5% (95% confidence interval = 12.8-27.6) decrease for weekly respiratory ED visits and an immediate 41.2% (95% confidence interval = 14.4-59.9) decrease in ED visits for pediatric asthma, followed by an additional 4% per-month longer term downward trend. Longer term reductions, as compared with preclosure trends, were also observed for hospitalizations for chronic obstructive pulmonary disease. Conclusions: Our study provides strong confirmation that reductions in fossil-fuel-related air pollution produce both short-term and longer term respiratory health benefits.

Keywords: COPD; air pollution; asthma; interrupted time series analysis.

Figure 1.

Flow diagram of overall study design. Dashed boxes indicate the causal inference criteria addressed by the before versus after and/or exposure versus control comparisons. COPD = chronic obstructive pulmonary disease; ED = emergency department; PADOH = Pennsylvania Department of Health; PHC4 = Pennsylvania Health Care Cost Containment Council.

Figure 2.

Zip codes of the study populations. (A) Exposure site (near-Shenango coke plant; Avalon): 15202, 15225, 15233, 15212, and 15136. (B) Exposed positive control site (Near still-operating Clairton coke plant; Clairton): 15025, 15037, 15088, 15133, 15122, and 15045. (C) Unexposed negative control site (Urban area, no coke plant nearby; Johnstown): 15902, 15901, and 15905. Green dots indicate the Allegheny County Health Department air quality monitoring sites; blue dots indicate the U.S. Environmental Protection Agency Chemical Speciation Network air quality monitor sites. Adapted by permission from Reference .

Figure 3.

Trends of annual hospitalizations at study sites, 2013–2018, for (A) COPD and allied conditions, (B) asthma, and (C) respiratory symptoms, at the exposure (Avalon/Shenango) site, the exposed control (Clairton) site, and the unexposed control (Johnstown) site. The models were adjusted for seasonality. Blue solid line indicates the quarterly hospitalization rate before Shenango shutdown. Purple dashed line indicates the predicted trend of hospitalization rate without Shenango shutdown. Red solid line indicates the actual visit rate after intervention. The vertical line represents the day of the coke plant closure. COPD = chronic obstructive pulmonary disease.

Figure 4.

Segmented regression results for weekly respiratory ED visit rates at the exposure (Avalon) site, the exposed positive control (Clairton) site, and the unexposed negative control (Johnstown) site, 2013–2018. Blue solid line indicates the visit rate before Shenango shutdown. Purple dashed line indicates the predicted trend of visit rate without Shenango shutdown. Red solid line indicates the actual visit rate after intervention. The vertical line represents the day of the coke plant closure. ED = emergency department.

Figure 5.

Segmented regression results for monthly pediatric asthma ED visit rates at the exposure (Avalon) site, the exposed control (Clairton) site, and the unexposed negative control (Johnstown) site, 2013–2018. Blue solid line indicates the visit rate before Shenango shutdown. Purple dashed line indicates the predicted trend of visit rate without Shenango shutdown. Red solid line indicates the actual visit rate after intervention. The vertical dashed line indicates the day of the Shenango coke plant closure. ED = emergency department.

Figure 6.

Summary of study findings. White header row and column cells refer to the respiratory health benefits tested at the near-Shenango exposure site. Gray header row and column cells refer to internal and external controls. Green boxes refer to the statistically significant decrease observed. Blue boxes indicate that no significant change was observed. Red boxes indicate that a statistically significant increase was observed. COPD = chronic obstructive pulmonary disease; ED = emergency department.