This doc is a compilation of figures related to the investigation into a recycling facility in Olneyville. The site is suspected of smelting lead (evidence: lead found on site and in wastewater downstream but not upstream of the site) and illegally operating certain furnaces (evidence: a video of black smoke and flame coming out of the chimney at the site). The company denies both allegations.
Child BLL level investigation
The first step in this investigation was to examine the blood lead levels of children near the site. For children who had multiple tests, the maximum result was used and other test results were excluded from analysis. a 1,500 foot radius was drawn around the site and used to subset “nearby” children. This was compared against the statewide population as well as a subset of results in all census tracts touching the census tract the site is contained in. This second comparison was included because Providence is known to have higher BLL than the non-urban-core parts of RI, however Providence itself is heterogeneous in population distribution so this method ensured it would capture a similar population.
In the following table, we condsider the mean BLL, the percent of children exceeding 5 µg/dL (which was the standard until recently), and the percent of children exceeding 10 µg/dL by each aforementioned geography. Counts by geography are also included.
| name | n | mean BLL | percent above 5 µg/dL | percent above 10 µg/dL |
|---|---|---|---|---|
| radius | 333 | 2.186 | 5.706 | 2.402 |
| tracts | 7549 | 2.307 | 7.074 | 2.119 |
| statewide | 116883 | 1.715 | 4.250 | 1.328 |
As seen in the table above, the area around the site is different than the statewide average (mean difference p-value < 0.001). However, these levels were not significantly different than those in surrounding neighborhoods (mean difference p-value = 0.406).
Local vs. Citywide PM 2.5 Concentration Investigation
Our next step in the investigation was to examine trends in air quality around the site. Dr. Hastings has installed 22 air quality monitors across Providence, including one approximately a block from the site being investigated.
Below is the plot of the nearby site (United Way) vs. the citywide mean. The trends are what are expected for AQM data - a non-normal distribution of relatively low baseline levels interspersed with “spikes” of high AQM which significantly impact mean values.
While the above plot is informative, it is not particularly useful for comparing between the sites. As such, a difference plot was created (United way PM 2.5 minus the city mean PM 2.5 values). As the following plot shows, there were a number of times when the United way site exceeded the city mean value, however those exceedances were significantly larger and more frequent in the first half of this time period, up until mid-March. In the following plot, please note that the horizontal dashed line represents the 95% confidence interval for the difference between United way and the city mean. Note that every 95% CI positive exceedance happened in the first half of this period.
An obvious next step was to investigate the conditions of these days where the difference was statistically significant. The majority of these exceedances occurred when the city mean values were low.It is important to note that the majority of AQM reading s are low, however these are also the days when a single point source contributor will be most likely to be detectable due to lack of confounding inputs.
We also examined the times that these exceedances were occurring. All exceedances occurred during working hours, between 8 AM and 4 PM (1600 in military time). The facility in question’s operating hours are 8 AM to 4:30 PM.
Next Steps for the Investigation
We have requested soil lead samples for the surrounding area to compare against BLL sample results. This will allow us to compare whether the exposure - outcome trends are consistent between populations.
We will be working with DEM to determine if the secession of these “larger” exceedances aligns with then the facility realized it was under investigation.
We will work with DEM to identify any complaints filed during the period of data collection, and it they have time and date information we will compare those days to the data to identify if they co-occur.