Community Engaged Cumulative Risk Assessment of Exposure to Inorganic Well Water Contaminants, Crow Reservation, Montana

An estimated 11 million people in the US have home wells with unsafe levels of hazardous metals and nitrate. The national scope of the health risk from consuming this water has not been assessed as home wells are largely unregulated and data on well water treatment and consumption are lacking. Here, we assessed health risks from consumption of contaminated well water on the Crow Reservation by conducting a community-engaged, cumulative risk assessment. Well water testing, surveys and interviews were used to collect data on contaminant concentrations, water treatment methods, well water consumption, and well and septic system protection and maintenance practices. Additive Hazard Index calculations show that the water in more than 39% of wells is unsafe due to uranium, manganese, nitrate, zinc and/or arsenic. Most families’ financial resources are limited, and 95% of participants do not employ water treatment technologies. Despite widespread high total dissolved solids, poor taste and odor, 80% of families consume their well water. Lack of environmental health literacy about well water safety, pre-existing health conditions and limited environmental enforcement also contribute to vulnerability. Ensuring access to safe drinking water and providing accompanying education are urgent public health priorities for Crow and other rural US families with low environmental health literacy and limited financial resources.

. Averages and standard deviations of Mn concentrations in mg/L, in home well water by ZIP code, on the Crow Reservation.

Nitrate
The 4.3 percent of wells exceeding the EPA MCL of 10 mg/L for nitrate plus nitrite as N ("NO3 -") is comparable to the 5% of wells identified by the IHS as exceeding the MCL (Table 4). Elevated NO3 -in well water was found primarily in the Bighorn River Valley, where irrigated agriculture is most extensive ( Figures S3 and S4).

Arsenic
Arsenic occurs primarily in Bighorn and Little Bighorn River valley wells. There are two EPA standards for As, an MCLG of 0.000 mg/L and an MCL of 0.010 mg/L. Twenty-seven percent of home wells had detectable As, i.e. they exceeded the MCLG, while only 1% exceeded the MCL (Table 4).

Uranium
There are two EPA standards for uranium: a maximum contaminant level goal (MCLG) of 0.000 mg/L, and a maximum contaminant level (MCL) of 0.030 mg/L. More than two thirds of wells tested positive for uranium, i.e. exceeded the MCLG (Table 4), and hence consuming this water incurs a low-level health risk. Uranium exceeding 0.030 mg/L was found in 6.2% of wells (Table 4), rendering the water from these wells unsafe for consumption. IHS data on uranium in home well water are lacking for the Crow Reservation.
Spatial analysis found that residents in the lower Bighorn River valley are most at risk: wells in the St. Xavier ZIP code average 0.018 ± 0.030 mg/L and those in the Hardin ZIP code average 0.010 ± 0.021 mg/L U. In the Crow Agency and Wyola ZIP codes of the Little Bighorn River valley, few wells exceeded the MCL, but many have elevated U concentrations. See Eggers et al. [99] for further information and a map.

Geomasking
While representation of point features is important for visualizing local patterns of environmental and health data, geovisualization of such information may increase the risk of participant identification because reverse geocoding methods could be used to locate study participants with derived latitude and longitude coordinates [127]. Geomasking methods have been developed in response to this risk so that data visualizations can be employed while protecting participant confidentiality [126]. Some masking methods, such as data aggregation, have low data resolution and decreased power to detect spatial patterns when the phenomenon of interest is local or crosses administrative boundaries [128]. Other classes of geomasking methods do not aggregate data and rely on random perturbation, which randomly shifts the point feature to a new location. Here a random perturbation geomasking method known as the donut method is employed, which shifts a point to a location between a minimum and maximum distance threshold and in a random direction. This method commonly scales the distance threshold based on population density of the study area, with larger scaling in lower population areas [129].
For the visualizations created for the present study each point location was shifted a random direction and located in an area between a minimum distance (Dmin) and maximum distance (Dmax) from the original location: In equations (S1) and (S2) pd is the population density of the administrative district in which the point is located, a and b are predefined integers that ensure the point is moved a minimum distance, and x and y are random integers that increase the shifting distance. The minimum and maximum distance thresholds were proportional to the inverse of population density (pd) multiplied by a predefined minimum movement distance (a or b) plus a random distance integer (x or y) [126]. Population density was determined by spatially joining the US Census 2010 Census Block Group for each point; the range of Dmin and Dmax values are not reported here to maintain confidentiality. In ArcGIS (v 10.3.1; ESRI, Redland, CA, USA) the geomasking was executed using a Model Builder script. The Buffer Tool was used to generate geodesic circular buffers using Dmin and Dmax for each point. Subsequently, the minimum buffer was erased from the maximum buffer leaving a ring that defined the area of possible relocation ( Figure S5). The Create Random Points Tool was used to generate one new location within each buffer ring and a spatial join was used to move the original attributes to the new location feature for visualization. Figure S5. Illustration of the donut geomasking method including an example buffer ring and a hypothetical original and shifted point location.