On 16 October 1996, a malfunction at the Swan Hills Special Waste Treatment Center (SHSWTC) in Alberta, Canada, released an undetermined quantity of persistent organic pollutants (POPs) into the atmosphere, including polychlorinated biphenyls, dioxins, and furans. The circumstances of exposure are detailed in Part 1, Background and Policy Issues. An ecologically based, staged health risk assessment was conducted in two parts with two levels of government as sponsors. The first, called the Swan Hills Study, is described in Part 2. A subsequent evaluation, described here in Part 3, was undertaken by Health Canada and focused exclusively on Aboriginal residents in three communities living near the lake, downwind, and downstream of the SHSWTC of the area. It was designed to isolate effects on members living a more traditional Aboriginal lifestyle. Aboriginal communities place great cultural emphasis on access to traditional lands and derive both cultural and health benefits from “country foods” such as venison (deer meat) and local fish. The suspicion of contamination of traditional lands and the food supply made risk management exceptionally difficult in this situation. The conclusion of both the Swan Hills and Lesser Slave Lake studies was that although POPs had entered the ecosystem, no effect could be demonstrated on human exposure or health outcome attributable to the incident. However, the value of this case study is in the detail of the process, not the ultimate dimensions of risk. The findings of the Lesser Slave Lake Study have not been published previously and are incomplete.
Persons living traditional lifestyles, such as Aboriginal communities in northern Canada, might be at higher risk from certain types of pollution than urban residents; the risk is specific to the situation.
In the real world, studies in which decisions are necessarily based are not always complete or performed as they might be for research studies.
Confounding factors, such as lifestyle and cigarette smoking, complicate interpretation of human risk assessment.
Incidents in which people are exposed to toxic substances never occur in a social vacuum. Certain communities are invariably more exposed or at risk than others. In this part of the case study, attention was focused on the Aboriginal communities of west-central Alberta south of Slave Lake and downstream from Swan Hills following an incident of the release of persistent organic pollutants (POPs).
In 1996, an incident at the Swan Hills Special Waste Treatment Center (SHSWTC, now called the “Swan Hills Treatment Facility”) in the town of Swan Hills, province of Alberta, resulted in uncontrolled emissions of polychlorinated biphenyls (PCBs), polychlorinated dibenzo-dioxins (PCDDs), polychlorinated dibenzofurans (PCDFs) and other organochlorines . The background and history of this incident is provided in Part 1 of this case study . Human health risks associated with environmental release of POPs can be inferred from the ecological behavior of the compounds and pathways by which the POPs reach human populations. The ecotoxicology (fate and disposition) of POPs released during this event is described in Part 2 .
ABORIGINAL HUMAN HEALTH RISK STUDIES
Part 3 of this case study describes the exposure, as represented by contaminant levels determined in the Aboriginal population, presumed vulnerability. This vulnerability could draw equally from closer proximity to natural pathways of exposure, higher consumption of “country foods” from living off the land, and cultural factors in food preferences and the social meaning of food restrictions and contamination. Health risk is a central concern of Aboriginal cultures and their traditional knowledge. It is also a critical sticking point in relations between band governance and government .
The Aboriginal population places great cultural emphasis on access to traditional lands and country foods such as venison (deer meat) and local fish. The Aboriginal population is diverse and fluid, although status is determined by registration as part of a local community (band). In this part of Alberta, the aboriginal communities consist primarily of Woodland Cree and Métis peoples, but mixed family units and individuals of other nations are commonly present (such as Objibwa or Chipweyan or Métis). The Métis people are descendants of Europeans (French or broadly Scottish and often specifically Orkney Islander) and aboriginal peoples (particularly Cree and Ojibwa) with constitutional recognition, and a distinct and (in Alberta) often but not always francophone culture. Aboriginal populations in rural or remote Alberta, as in other parts of Canada, rely on hunting, fishing, and gathering, for sustenance (seldom exclusively), cultural tradition, and political expression (sometimes following the movement called “decolonialization”).
This case study also illustrates the difficulties of adapting a risk management strategy to groups with different cultural values and expectations. Community and individual responses to such incidents, and the development of health advisory messages, may depend on presenting information on exposure and risk in terms consistent with cultural patterns among subpopulations in the community.
Names of provincial agencies are given as they were in 1997; most have been changed since, but the areas of responsibility are obvious from their names.
THE LESSER SLAVE LAKE STUDY
In Canada, the provincial governments have primary responsibility for evaluating both environmental and human health risk, with some exceptions. The role of the Government of Canada is constitutionally circumscribed but the federal government has responsibility for First Nations populations. This generally takes the form of paying for provincial health insurance premiums for First Nations peoples and providing public health services either directly or reimbursing provincial agencies. Health Canada became directly involved in investigating Swan Hills in July 1997 at the request of the Lesser Slave Lake Indian Regional Council, which wanted a focused and more detailed analysis of their situation than the Swan Hills Study could provide.
A special study was designed to achieve the following objectives:
To assess the environmental exposure and risk of the Lesser Slave Lake First Nation populations, specifically, to PCBs, dioxin/furans and several other contaminants (cadmium, lead, and mercury);
To establish the traditional consumption patterns and uses of freshwater fish and wild game of Aboriginal people in the Swan Hills Region (not reported here);
To reexamine the Alberta Consumption Advisory for deer and moose meat in the context of First Nations consumption information (not reported here);
To determine the association, if any, between traditional food consumption and contaminant burdens (not reported here);
To investigate any health effects of environmental exposure to contaminants and their source (not reported here);
To provide the Lesser Slave Lake First Nation and political authorities with contaminant exposure and consumption information (not reported here) specific to their population to help them advise their population and manage health issues.
The Lesser Slave Lake Health Study, like the broader Swan Hills study conducted by Alberta Health described in Part 2, was designed primarily as an exposure assessment. The Study was divided into three phases. Phase I (1997) examined a subset of Aboriginal peoples resident in the area who were determined, on lifestyle and cultural grounds, to be most likely in the community to consume large quantities of country foods and therefore to experience a higher exposure from that source. Phase II (1997), not described in detail here, examined a randomly selected sample of Aboriginal residents in the area, using largely the same methodology. Phase III (2002) re-examined volunteers among the Phase I sample to determine change over five years.
Phase I: Study Design and Quality Assurance
Phase I of the study was an assessment of a primary risk group consisting of 50 First Nations people aged 15 years and older selected by band councils from among eight communities initially participating in the study. These individuals were believed to be at higher risk of exposure because they were recognized by their communities as living traditional lifestyles and engaged in heavy consumption of country foods. They were residents of four communities, A, B, C, as in the Swan Hills Study, and a fourth location here called D. (Aboriginal communities are not identified, in order to minimize harm.) The populations number in the hundreds each but none counts as many as 1,000 residents.
Residents were studied with serum levels reported for organochlorine compounds, blood cadmium, and lead levels; hair analysis for mercury; and a survey questionnaire covering health, eating patterns, and practice of a traditional lifestyle. The SHSWTC was not suspected of emitting metals at the time of the incident and metal-containing solutions were not part of its intake. Their inclusion, together with lead, probably represented a target of opportunity to perform health screening.
Analyses were conducted by various laboratories with elaborate quality assurance, including samples comparing Health Canada laboratory results with those from research and commercial laboratories and from Environment Canada. In general, with many exceptions, congener-specific levels were close (±22% on average). Mean levels of dioxins and furans were less close (±52%).
Phase II: Study Design and Quality Assurance
Phase II of the Lesser Slave Lake Study is not presented in detail because the findings essentially duplicated that of Phase I. Phase II was an effort to obtain a larger and more diverse population sample to determine the generalizability of findings and to conduct a health symptoms survey.
Phase II was essentially a convenient sample stratified to represent the population of interest in the absence of a true random sample. Selection of subjects could not be randomized because there were far too few individuals, the use of band intermediaries precluded randomization (and even strict confidentiality), and the status was ambiguous in many cases. The status (status as enrolled in a band registry, non-status, Métis) was not considered a major concern because it was not correlated with participation in traditional lifestyle; all Aboriginal community residents participated at least to some degree in traditional practices.
The health symptom reports are difficult to interpret because of obvious methodological lapses and lack of relevant comparison data. Rather than present incomplete and possibly flawed information, the findings are not reported here.
Phase III: Study Design and Quality Assurance
Blood was taken from all participants and serum was analyzed for PCBs, PCDDs and PCDDs, whole blood for cadmium and lead. Hair samples were taken and analyzed for mercury. Values presented in this report are those reported from the laboratory of Health Canada, which applied the same quality assurance standards as in Phase I. Results were interpreted with reference to Canadian and international comparison populations [5–7].
Because the biomonitoring determinations were not paired, but are drawn from a subset of the same population, the results of the Phase III (follow-up) study should be treated as a new determination on a second sample of the Phase I participant population rather than as either a completely independent determination or as a repeated measure on the same group of participants. Quality assurance procedures followed in Phase III were the same as in Phase I except that there was no inter-laboratory comparison for PCB congeners.
The survey questionnaire that was initially part of Phase I was not repeated in the Phase III (follow-up) study. It was assumed that health status, eating patterns, and practice of a traditional lifestyle would not have changed substantially in the meantime for subjects.
Participants in the Phase III (follow-up) study were also asked to rate their personal health on a five-point scale.
The findings from the Lesser Slave Lake Study were first presented in detail in 2003 to the affected First Nation communities and conveyed in a report to Health Canada. The results have not been published previously. (These results cannot be assumed to apply today to the same population.)
Among the 50 original subjects surveyed in 1997, 31 (62%) were males older than 18 years, 10 (20%) were females aged 15 to 45 years, and 9 (18%) were females aged 46 years or more. The average age (47.7 years) suggests an older sample than is likely to be representative of the population and is consistent with the long duration of residence in the community (37.6 years). Cigarette smoking was prevalent (62% active, 66% passive) but it was not reported how this and other characteristics were distributed among the three groups of interest.
Of the original 50 participants chosen for traditional lifestyles and presumed high exposure, 41 participants, 24 men and 17 women, agreed to be retested for Phase III. The average age of the men was about 48.4 years (with three missing birth years) and of the women was about 48.2 years (with one missing birth year). Participants had lived in the community for a minimum of 25 years and a maximum of 82 years (mean 52.41 years) confirming that most respondents were lifelong or nearly lifelong residents of their community. Aboriginal status was distributed with 36 (88%) declaring registered First Nation status, 3 (7%) declaring Métis identity, and the rest being unknown.
Because of the terms of the study agreement with the participating First Nations that ensure strict confidentiality, it was not possible to compare paired biomonitoring results for the same individuals in Phase I and Phase III or between the Lesser Slave Lake Study and the Swan Hills Study.
Phase III (Follow-Up) Study
Cigarette smoking remained highly prevalent in the Aboriginal communities: 66% of subjects smoked and 73.2% lived with at least one household member who smoked. The high prevalence of smoking is a particular issue in the interpretation of cadmium levels, as will be discussed.
The results of the repeated biomonitoring performed in 1997 and 2002 are summarized in Table 1.
|Analyte .||Mean in 1997 .||Mean in 2002 .|
|PCBs as Arachlor 1254 (ppb), expressed as Toxic Equivalency (TEQ) |
|PCBs as Arachlor 1260 (ppb), expressed as Toxic Equivalency (TEQ) |
|Dioxins/Furans (pg/g = ppt)|
|Analyte .||Mean in 1997 .||Mean in 2002 .|
|PCBs as Arachlor 1254 (ppb), expressed as Toxic Equivalency (TEQ) |
|PCBs as Arachlor 1260 (ppb), expressed as Toxic Equivalency (TEQ) |
|Dioxins/Furans (pg/g = ppt)|
*Detailed statistical analysis was not provided.
Briefly, all measures save one are less than the initial Phase I values. No measure, however, was statistically significantly less than the Phase I value at p = 0.05. Taken together, however, the uniform fall in all contaminant levels is highly significant, at least at p < 0.0001 (nonparametric trend, counting the two Arachlor formulations as if independent).
The one exception is cadmium, for which the 2002 Phase III (follow-up) testing round showed an average blood level for men of 0.335 μg/dl compared to 0.3030 in 2002, the sole example of an unchanged value, being within measurement error.
Risk determinations based on consumption patterns used for the original Phase I and Phase II studies could not be reconstructed for the Phase III (follow-up) study because the algorithm was not available.
Data on body weight and height were collected in the Phase III (follow-up) study, using both metric and Imperial units. (Confusion between the two was apparent in some of the early data sheets.) Among men, 18 subjects reported weight, always in pounds, averaging 226.3 pounds (range 143.0 to 397.0) or 102.8 kg. Five women reported weight averaging 188.8 pounds (range 160.0 to 211.0) or 85.8 kg. The height data, however, are obviously invalid and therefore not reported. (Indicators of bias include a range of height for 28 respondents that included values to 1 and 11 inches and a finding that 28 respondents were recorded as exactly 5 feet tall.)
Only 11 of the 24 men and 3 of the women rated their health as “Excellent” or “Very Good” and 7 each rated their health as only “Fair” or “Poor”. Although the numbers are small and relevant comparisons are lacking, it would appear that this community does not view itself as being in particularly robust health.
Participants were also asked to evaluate the perceived risk of PCBs to personal (i.e. “your”) health, on a scale from 1 to 4, where 1 = no risk, 2 = minor risk, 3 = moderate risk, and 4 = major risk. With 31 responses, the mean rating on the scale was 2.87 (S.D. 0.99). Fewer subjects responded to this item on the questionnaire, but 13 of the 18 men and 9 of the 13 women responded that PCBs were a major or moderate risk to health. This high level of concern may reflect the intensity of the study itself, which from the point of view of the participants cannot be easily rationalized unless there were serious concerns with respect to their health.
There were many methodological issues with this part of the Lesser Slave Lake Study, with obvious problems of data entry, confusion over metric and Imperial measurement, data gaps, and imprecise categorization of subjects as living traditional and nontraditional lifestyles. (This turned out to be a false dichotomy, since all subjects had some elements of a traditional lifestyle and none were living strictly as their forebears did in 1899, when Treaty 8 was signed.) The wording of the questionnaire was also problematic, because it did not conform to standard instruments and so could not be reliably compared.
Most difficult, however, is that the biomonitoring data, while intended to be repeated measures of the same individuals actually reflect differently constituted groups. It is clear that the Phase III (follow-up) population differs in some important ways from the original Phase I population. As expected, they were older, on average about five years, at the time of the original Phase I study. They may differ in attitudes or accessibility, by being available to investigators.
The interpretation of biomonitoring data would be more difficult if there were evidence for elevations in contaminant levels, given the methodological issues and missing information. In general, levels of POPs (expressed in ng/kg, or parts per trillion) found in blood in all phases of the Lesser Slave Lake Study were very low compared to studies in other populations, Health Canada Guidelines, and compared to residents of the community of Swan Hills (see Part 2). The predominant findings are therefore a “cold negative”, in the sense that they do not suggest that a toxicologically significant level of exposure has occurred in this population. However, there are anomalies that require attention.
Briefly, all measures of POPs save one are less than the initial Phase I values. Measurements of total PCB, dioxins/furans, and DDT/DDE are less in the Phase III (follow-up) study by as much as 50% in five years (for Arachlor 1260 in men). The trend toward lower values is consistent across biomonitoring measures for POPs. No measure, however, was statistically significantly less than the Phase I value (at p = 0.05). Taken together, however, the uniform fall in all contaminant levels is highly significant, at least at p < 0.0001 (nonparametric trend, taking each sex as an independent replication and counting the two Arachlor formulations as if independent). Measures of total PCB are less in the Phase III (follow-up) study by as much as 50% in five years (for Arachlor 1260 in men). This greatly exceeds the rate at which these substances are thought to be metabolized. This observation suggests the presence of bias in the reported results, most likely that the populations are not comparable.
There are possible various explanations for this apparent trend (termed apparent because the two cross-sectional samples are not from exactly the same subjects):
The subset of participants in the Phase III (follow-up) study may be unrepresentative of the original Phase I participant group (clearly younger) and less exposed to contaminants of interest.
Laboratory analytical differences or technique may have systematically lowered the reported value of contaminants in the Phase III (follow-up) study.
The levels of POPs and lead in body fluids may have dropped over the time period and the fall for all Phase I participants may be reflected in this subsample.
Lifestyle changes, possibly adopted in response to concern over toxicity, may have reduced exposure to contaminants across the board, except for cigarette smoking.
Although the first explanation cannot be ruled out, it seems unlikely, given the consistency of the lower levels among both sexes and all POPs contaminants. As well, the participants in Phase I, from which the participants in phase III were drawn, were selected because they were known to engage in a more traditional lifestyle and were therefore more likely to be exposed by ecological pathways. Thus, their exposure would be expected to be higher than for other residents of the community. An elapsed time of 5 years would be expected to be associated with higher and rising levels in 2002 both because of cumulative exposure and aging, which is associated with increased body burden of POPs over a lifetime. However, in the event the exposure indicators appear to be relatively uniform, low, and possibly falling (bearing in mind that these are not strictly repeated measures). The third explanation is possible but cannot be proven in the absence of repeated measures in the same individuals. The fourth explanation is likely to be a contributing factor but would not be expected to result in such a large reduction. Thus, there is no single satisfactory explanation for why POPs levels in Phase III are so substantially lower than in Phase I.
Lead levels are at levels expected for adults, especially given that this population often shoots and some may cast their own bullets. The most likely reasons that lead levels were lower in 2002 than in 1997 are either that fewer individuals are pouring and shooting their own home-made ammunition or that those who do are taking greater precautions with lead exposure.
The levels of cadmium and mercury are sufficiently low that public health risk is very unlikely. Chrystina Lake has been identified as having mercury levels that are moderately high compared to other bodies of water but do not exceed the Canada Health guidelines . The stable cadmium levels are readily explained because the driving exposure for cadmium in this population is clearly cigarette smoking, not environmental sources in this population. (Cadmium is a particular problem in Aboriginal communities where caribou is harvested.)
Sometimes, predicted consequences do occur and other times they do not. The rationale of both the Swan Hills and Lesser Slave Lake studies was that POPs had entered the ecosystem and pathways that would eventually result in human exposure. However, no effect on human exposure was demonstrable although sufficient time elapsed since the incident. The magnitude of exposure did not appear to constitute a major risk to health.
Ecosystem fate and mobility studies (ecotoxicology) can be used to anticipate future human health risk, as in the Swan Hills Study. A survey of contaminant exposure is limited by the toxicokinetics of the contaminant under consideration. In the case of POPs, by definition, the issue of concern is cumulative exposure and long-term consequences. The Lesser Slave Lake Study was methodologically problematic but sufficient for the purpose, which was to determine public health risk and to target any interventions that would be necessary operationally.
An ideal study would have integrated the Swan Hills Study and the Lesser Slave Lake Study, using standardized methods. In the real world, studies on which decisions are necessarily based are not always complete or performed as rigorously as they might be for research studies. As long as the information is actionable and reliable, that does not diminish their importance or utility for risk management but it does diminish their generalizability to other situations. Public health agencies must prioritize public protection first, even at the expense of research when resources are limited.
Many scientists and public health professionals took part in the several studies that together constituted the evaluation of the Swan Hills incident. Data cited in this case study are publicly available with the exception of the Lesser Slave Lake study, which have been publicly presented but is published here in the accessible journal literature for the first time. The case study was taught for several years as a case study in PubH 243 Public Health Practice at the George Washington University. Ms. Amy O’Connor developed the narrative further for her MPH research project in 2007.
Funding for Dr. Guidotti’s participation in the Lesser Slave Lake Study was limited to interpretation of data provided by Health Canada and formal presentation of the findings to elders of the First Nations communities involved at a meeting convened for the purpose in 2003.
The author has no competing interest to declare.
Swan Hills: Questions for Discussion
During the events described in this case study, the author served as a principal consultant in environmental health for Alberta Environment, Alberta Health, and Alberta Justice, and, later, Health Canada. (Agency names are given as they were in 1997.)