Risk factors associated with perceptions of drinking water quality in rural Saskatchewan

Abstract

A cross-sectional study used anonymous postal questionnaires to investigate risk factors associated with perceptions of water quality and risk from drinking water in rural Saskatchewan. From the 7500 questionnaires mailed out, the 2065 returned responses were analyzed using generalized linear mixed models. Reporting a drinking water advisory significantly increased the likelihood of any aesthetic complaint with tap water. Using truck-delivered water, being older, being male and living in an area for more than 10 years significantly decreased the likelihood of an aesthetic complaint. Having an aesthetic complaint significantly increased the likelihood of believing that tap water is not safe to drink. However, using a community water supply decreased this likelihood. Reporting a water advisory significantly increased the likelihood of believing tap water was not safe, but the magnitude of the effect was greater for those who used a ground water source than those who did not. Using a ground water source significantly decreased the likelihood of believing the tap water was not safe, but only for those who never had a water advisory.

Une étude transversale utilisait des questionnaires anonymes envoyés par poste pour étudier les risques associés à la perception de la qualité de l’eau et des risques de l’eau potable dans la Saskatchewan rurale. Au total, 7500 questionnaires étaient envoyés par poste, et 2065 réponses étaient analysées en utilisant des modèles linéaires généralisés mixtes. Signaler un avis concernant l’eau potable augmentait la probabilité d’une plainte esthétique avec de l’eau du robinet. Utiliser l’eau livrée par camion, d’être plus âgé, être un homme et vivant dans une zone depuis plus de dix ans ont tous diminuait la probabilité d’une plainte esthétique. Avoir une plainte esthétique augmentait la probabilité de croire que l’eau du robinet n’était pas potable. Cependant, utilisant l’approvisionnement en eau de la collectivité diminuait cette probabilité. Signalement d’un avis concernant l’eau potable augmentait la probabilité de croire que l’eau du robinet n’était pas sûre, mais l’ampleur de l’effet était plus importante pour ceux qui utilisaient une source d’eau souterraine que ceux qui n’avaient pas. Utilisant une source d’eau souterraine diminuait la probabilité de croire que l’eau du robinet n’était pas sûre, mais seulement pour ceux qui n’avaient jamais eu un avis concernant l’eau potable. Les résultats soulignent la complexité des facteurs impliqués dans la qualité de l’eau et la perception des risques, en particulier dans les zones rurales où les caractéristiques des sources d’eau sont variées.

Introduction

Perceptions of drinking water quality and risk are influenced by a complex set of factors relating to sensory perception, risk tolerance and social, psychological and economic factors. In the context of drinking water, risk perception has been defined as “an individual’s subjective judgment (based on aesthetic and non-aesthetic qualities) about drinking water” (Anadu and Harding 2000, 82). Risk perception largely depends on intuitive judgment and is also influenced by cultural and social factors (Slovic 1987), including race, gender and socioeconomic factors (Finucane et al. 2000; Dosman et al. 2001).

In a review of the factors related to water quality and risk perception, Doria (2010) identified several factors that influence these perceptions, including aesthetic qualities, prior experience, familiarity, interpersonal information, trust and control issues, and personal and cultural influences. Studies of water quality and risk perception in Canada and elsewhere have shown that the aesthetic qualities of drinking water, especially taste and odor, are associated with quality and risk perception (Jones et al. 2007; Doria et al. 2009). However, many potentially harmful water contaminants are not expected to change the aesthetic qualities of water, and aesthetic characteristics can be unreliable for making risk judgments (Turgeon et al. 2004). Nevertheless, it has been suggested that personal experiences typically have the largest impact on perceptions of drinking water risk and quality. The aesthetic qualities of the water represent the main experience most consumers have with their water quality and are, therefore, the most readily available metric for making judgments (Doria 2010).

A sense of control over a hazard could reduce the risk perceived (Doria 2010). Residents with a private water supply could feel they have greater control over their water supply and therefore have more confidence in its safety. Trust in the water supplier, relevant for those with community treated water supplies, has also been identified as important in the perception of water quality and risk (Doria et al. 2009; McSpirit and Reid 2011; Saylor et al. 2011), as has familiarity with the water supply (Dietrich 2006; Doria 2010). Demographic and socioeconomic factors have also been found to play a role in water quality perceptions, although their relative importance and effect vary depending on the community (Turgeon et al. 2004; Doria 2010; McSpirit and Reid 2011).

An understanding of the factors associated with water quality and risk perception is important for developing public health strategies that promote safe drinking water. This is particularly important in rural areas, where a number of water sources of varied quality might be employed (Corkal et al. 2011), private sources are common and effective dissemination of timely information and educational materials can be challenging. Furthermore, we are unaware of any large-scale studies of water quality and risk perceptions in rural Saskatchewan, where the types of water sources used can vary considerably.

Differences in survey methods used across communities can make comparing studies of water quality and risk perception challenging. Insufficient clarification of the dimensions of quality and risk pose an additional barrier to the comparison of factors associated with these perceptions. Although the concepts of quality and risk might be expected to be closely related, Janmaat (2007) suggested that concerns about quality and risk are somewhat separable, with concerns tending to be strongest in only one of these dimensions. Our primary objectives were to investigate risk factors associated with reporting any aesthetic complaints about the household’s tap water, and also with the perception that the tap water is not safe to drink. The secondary objectives were to examine risk factors associated with the fear that the household’s water source will become contaminated, and the perception that someone had become ill as a result of drinking the household’s tap water.

Materials and methods

Design

A postal questionnaire was sent to 7500 rural households in six regions of Saskatchewan in the fall of 2011. The questionnaire was anonymous and distributed through Canada Post’s Unaddressed AdMail service. Postal code geography files (Platinum Postal Code Suite 2006; DMTI Spatial Inc., Markham, ON) were used in conjunction with Canada Post Householder Counts to select postal codes to which the questionnaire was distributed. A commercial geographic information system (ArcMAP, ESRI, Redlands, CA) was used to calculate the smallest radius for each region that would be necessary to include centroids of eligible postal codes encompassing 1250 houses and farms. Postal codes with no farms were excluded, and for postal codes that contained more than 200 houses, the survey was sent only to farms to ensure that the distribution of the questionnaire would be primarily to rural households. Questionnaires were sent to 1250 households from between nine and 12 postal codes (median = 10) in each of the six regions, for a total of 60 postal codes. The resultant data were hierarchical with clustering by postal code nested within region.

The questionnaire was four pages long and consisted of questions about water sources, perceptions of water quality and risk, experiences with water advisories and drinking water choices. It was based on a questionnaire that had been piloted in a different community the previous year. We asked that one member of the household over the age of 18 fill out the questionnaire and return it in a postage-paid, pre-addressed envelope. Because the questionnaires were not addressed to specific households, distributing reminders was not practical. The informed consent of the participants was obtained, and the study protocol was approved by the Behavioral Research Ethics Board of the University of Saskatchewan.

Outcomes and risk factors

The primary outcomes of interest in this analysis were whether residents specified any aesthetic complaint about their household’s tap water (odor, bad taste, discoloration or cloudiness), and whether residents perceived that their household tap water was not safe to drink. Fear that the household’s water source will become contaminated and the perception that someone had become ill from drinking the household’s tap water were also modeled as outcomes.

Reporting an aesthetic complaint was a dichotomous variable created from a question which allowed multiple responses regarding characteristics of the household’s tap water. If any of odor, bad taste, discoloration or cloudiness were selected, the aesthetic complaint variable was coded as “yes,” and if none were selected, the variable was coded as “no.” The frequencies of the individual aesthetic complaints were also calculated, and the agreement among pairwise combinations was estimated using Kappa coefficients (κ).

A causal diagram (Figure 1) was constructed to help guide model development. The risk factors included a variety of measures related to household tap water sources: use of a community treated water supply, use of a private water supply, water delivered by truck, use of a public water station, whether the household had a cistern and whether the water source was ground water or surface water. In rural areas, households can use more than one water source; therefore, each of the possible types of water supply was analyzed separately. Water sources were not mutually exclusive.

Figure 1. Causal diagram used to guide model development for each of the outcomes.

History of having had any type of drinking water advisory, past or present, in the current household was assessed as a risk factor. As well, whether the home was in a town, length of time residing in the current community, age, gender and whether there were children under 18 years old residing in the household were also analyzed as risk factors.

Additionally, because aesthetic characteristics have been linked to drinking water risk perception in the literature, having an aesthetic complaint was included as a risk factor in the models both for the perception that the household tap water is not safe to drink and for the perception that someone has become ill from drinking the tap water.

Six age categories were recorded on the questionnaire. The three youngest age groups were collapsed into a single category due to low numbers of responses in these categories, so that only four categories were used in the analysis. Four possible categories for the number of years residing in the community were also collapsed into two categories for analysis.

Statistical analysis

Each outcome was modeled using a generalized linear mixed model (GLMM), specifying a binomial distribution and logit link function. When exploration of differences between groups is not the primary goal of the analysis, these models parsimoniously account for clustering by partitioning the overall variance in the data into variance components. Random intercepts were included for both postal code (pc) and region (reg) to account for any clustering in the data arising from the hierarchical structure of the data (Equation 1):(1)

Models were built for each outcome by first screening each risk factor individually in a logistic mixed effects model with random intercepts for postal code and region; any risk factor with a p-value < 0.2 was retained for consideration when building the final model. Prior to considering all remaining risk factors (p < 0.2), an intermediate model was used to identify the best subset of retained risk factors related to water source, keeping only those with a p-value < 0.05. Manual backwards selection was used to build the final main-effects model, retaining only risk factors with a p-value < 0.05. The last risk factor removed from the model was assessed for confounding based on whether its inclusion in the model led to a change greater than 10% in the regression coefficients for other risk factors. If age, gender and/or children in the home were not retained in the model but were unconditionally associated with the outcome, they were assessed for confounding by the same criteria. Biologically plausible interactions between risk factors retained in the final model were assessed at a 0.05 level of significance; in the case of categorical variables, a type 3 likelihood ratio test was used to determine if the interaction was significant. Akaike information criterion (AIC) values were used to assess competing models where there was doubt regarding the best model fit.

Models were built in Stata (StataCorp LP, College Station, Texas) with the xtmelogit command using a Laplacian approximation for efficiency. Using the risk factors identified in the model-building process, the final model parameters were estimated with the user-written Stata program gllamm (Rabe-Hesketh et al. 2004) using adaptive quadrature with 12 integration points. Population averaged probabilities were estimated using the gllapred marginal function (Rabe-Hesketh and Skrondal 2008).

Odds ratios were obtained by exponentiating the regression coefficients (e^β) for the risk factors included in each model. Odds ratios represent the relative odds of the occurrence of an outcome in those with a risk factor compared to those without (or with a different level of the risk factor). Odds ratios (OR) were reported along with corresponding 95% confidence intervals (CI).

The proportion of the variance accounted for by postal code (Equation 2) and region (Equation 3) was examined for each of the outcomes, using an approximation of the variance partition coefficient for the binomial outcome based on the latent response variable model (Browne et al. 2005): (2) (3)

Values were missing for all outcomes and risk factors from at least one survey. Any observations that were missing values for any of the risk factors or outcome for a given model were excluded from analyses including that variable. Therefore, the final number of observations used in each model varied and is reported for each model. Model assumptions were examined by evaluating the distribution of the residuals at each random effects level using Q–Q plots. Residuals were also examined for the presence of outliers and influential data points by plotting the standardized residuals at each of the random effects levels.

Results

Of the 7500 questionnaires mailed out, 2074 were returned. One was excluded because it was not filled out, one was discarded because it was received subsequent to data entry and analysis, and seven were excluded from the analysis because the postal code identifier had been removed from the questionnaire before returning the questionnaire. As a result, 2065 total observations were available for use in this analysis, for an effective response rate of 27.5%.

The median number of responses per region was 353 (range 327–368), while the median number of responses per postal code was 44 (range 2–108); frequencies of each outcome and risk factor were calculated (Table 1).

Table 1. Total numbers of complete and missing responses and the proportion of respondents at each level for each of the outcomes modeled and risk factors evaluated.

		Complete	Missing	Response	Frequency
		n	n		n	%^a
Outcomes
	Any aesthetic complaint about tap water	1984	81	Yes	501	25.3
				No	1483	74.8
	Believed tap water not safe to drink	1984	81	Yes	235	11.8
				No	1749	88.2
	Fear tap water will become contaminated	1988	77	Yes	706	35.5
				No	1282	64.5
	Someone had been ill from tap water	1784	281	Yes	57	3.2
				No	1727	96.8
Risk factors
	Private water supply	2059	6	Yes	1249	60.7
				No	810	39.3
	Community (treated) water supply	2059	6	Yes	640	31.1
				No	1419	68.9
	Water delivered by truck	2056	9	Yes	119	5.8
				No	1937	94.2
	Used a public water station	2059	6	Yes	121	5.9
				No	1938	94.1
	Had a cistern	2027	38	Yes	292	14.4
				No	1735	85.6
	Ground water source	1857	208	Yes	1349	72.6
				No	508	27.4
	Surface water source	1856	209	Yes	613	33.0
				No	1243	67.0
	Ever had a water advisory	1981	84	Yes	485	24.5
				No	1496	75.5
	Home is in a town	2047	18	Yes	525	25.6
				No	1522	74.4
	Gender	2005	60	Female	1053^b	52.5
				Male	952	47.5
	Age	2050	15	< 45 years	317^b	15.5
				45–54 years	446	21.8
				55–64 years	614	30.0
				≥ 65 years	673	32.8
	Number of years in community	2046	19	≤10 years	403^b	19.7
				> 10 years	1643	80.3
	Children reside in home	1932	133	Yes	437	22.6
				No	1495	77.4

^a Percentage based on number of complete observations.

^b Reference category.

The distributions of age categories and gender in our sample population were compared to data from the Canada 2011 Census of Population (Statistics Canada 2011) for the Census Subdivisions included in our survey regions (Table 2) to assess the representativeness of our sample. Awareness of issues such as agricultural runoff or pollution affecting the household water source was reported by 21.8% of respondents.

Table 2. Comparison of key demographic variables between the sample population and the Statistics Canada 2011 Census of Population for the rural census subdivisions included within the survey regions.

Category	Survey respondents^a		2011 Census of Population (%)
	N	(%)
Female	1053	52.5	47.3
Male	952	47.5	52.7
18–45 years	317	15.5	35.7
45–54 years	446	21.8	23.2
55–65 years	614	30.0	21.7
≥ 65 years	673	32.8	19.3

^a Total number of respondents: 2005 for gender and 2050 for age.

Having any aesthetic complaint

Overall, 501 (25.3%) respondents were dissatisfied with at least one of taste, odor, color or cloudiness of their tap water, and were classified as having any aesthetic complaint about their tap water. Of the respondents who knew the source of their tap water, the proportion with any aesthetic complaint was similar for those using ground water sources (23.8%) and those using surface water sources (24.0%). The frequencies of each type of complaint were also broken down by type of water source used (Table 3). The types of aesthetic complaint were not mutually exclusive and of the 501 respondents with any complaint, 236 (47.1%) specified more than one type of aesthetic complaint. The agreement between the various aesthetic concerns ranged from a Kappa statistic of 0.19 (between taste and cloudiness) to 0.48 (taste and odor) and indicated slight to moderate agreement between the various types of complaints over all respondents (Dohoo and Martin 2012). In the open comments area of the survey, 159 respondents (8.0%) wrote that their tap water had a high mineral content or hardness; of these, 141 (88.7%) used a ground water source.

Table 3. Proportion of respondents identifying each type of aesthetic complaint overall and by the type of water source used.

Complaint	All respondents (n = 1984)	By type of water source used
		Ground water (n = 1293)	Surface water (n = 592)	Private supply (n = 1194)	Community supply (n = 620)	Delivered by truck (n = 114)	Public water station (n = 119)
Odor	12.5%	10.5%	13.0%	12.3%	13.7%	7.9%	14.3%
Taste	14.5%	13.1%	13.7%	11.5%	20.3%	5.3%	19.3%
Discolored	11.2%	11.1%	10.1%	13.4%	7.7%	2.6%	13.4%
Cloudy	4.8%	3.5%	6.6%	4.8%	4.0%	2.6%	5.9%

After accounting for other significant risk factors (Table 4), respondents who had their water delivered by truck were less likely to have an aesthetic complaint (p = 0.02) than those who did not. Those who reported any type of water advisory were more likely to report an aesthetic concern (p < 0.001) (Figure 2) than those who had not experienced an advisory. Compared to the youngest age group, respondents in each of the older age categories were less likely to have an aesthetic complaint (p ≤ 0.006) (Figure 2). Similarly, those over 65 were less likely to report aesthetic concerns than those 45–54 (OR 0.7, 95% CI 0.5–1.0, p = 0.02). Being male (p = 0.005; Figure 2) and having lived in the community for longer than 10 years (p = 0.007) were also associated with a decreased likelihood of having an aesthetic complaint.

Table 4. Risk factors significantly associated with having any type of aesthetic complaint with the household’s tap water in the final multivariable model.

Risk Factor	OR	95% CI	p value
Trucked water delivery	0.5	0.3, 0.9	0.02
Ever had a water advisory	1.8	1.4, 2.3	< 0.001
Male	0.7	0.6, 0.9	0.005
Female	REF		< 0.001^a
Age
Age < 45	REF
Age 45-54	0.6	0.4, 0.9	0.006
Age 55-64	0.6	0.4, 0.8	< 0.001
Age 65+	0.4	0.3, 0.6	< 0.001
≤ 10 years in community	REF
>10 Years in community	0.7	0.5, 0.9	0.007
Random Effects	Variance	SE
Postal Code	0.140	0.073
Region	0.034	0.040
Number of observations = 1865

Note:

OR: odds ratio. CI: confidence interval. REF: reference category. SE: standard error.

^a Wald Type III test of fixed effect for age

Figure 2. Predicted probabilities of having any aesthetic complaint, comparing those who had a drinking water advisory to those who did not, for men and women in each age category averaged over the use of water delivered by truck and length of time residing in community. CI: confidence interval.

Because the effect of age on taste satisfaction varied by community in a previous study (Turgeon et al. 2004), we tested if the effect of age varied by region, and found no evidence that this was the case in our study.

The proportion of variance accounted for by postal code (4.1%) was greater than the proportion accounted for by region (1.0%). This represented a small change compared to the random effects from the null model (in which postal code accounted for 6.2% of the total variance, and region 1.0%), suggesting that the fixed effects did not account for much of the between-group variation in having an aesthetic complaint.

The perception that tap water is not safe to drink

After accounting for other risk factors (Table 5), those who reported having any aesthetic complaint about the tap water were more likely to agree that their tap water was not safe to drink compared to those who did not report any aesthetic complaints (p < 0.001; Figure 3). Respondents who used a community water supply were less likely to believe that their tap water was not safe when compared to not using a community supply (p < 0.001). An interaction was identified between the use of a ground water source and having experienced any type of drinking water advisory. Compared to those who did not use a ground water source, those who used a ground water source were less likely to believe that the household tap water was not safe to drink, but only when they had never had a drinking water advisory (p = 0.001; Figure 3). Compared to those who reported never having had a drinking water advisory, respondents who had experienced a drinking water advisory were more likely to believe that their tap water was not safe to drink, but the magnitude of this effect was greater for the respondents who used a ground water source (OR = 7.3, p < 0.001) than for those who did not use a ground water source (OR = 3.0, p < 0.001; Figure 3).

Table 5. Risk factors significantly associated, in the final multivariable model, with the perception that household tap water is not safe to drink.

Risk Factors		OR	95% CI			p value
Community treated water supply		0.2	0.1, 0.3			< 0.001
Any aesthetic complaint		4.6	3.3, 6.3			< 0.001
Ground water source x ever had advisory						0.02^a
	Use a ground water source compared to not using ground water for those who have ever had a water advisory	1.1	0.6, 2.0			0.77
	Use a ground water source compared to not using ground water for those who have not had a water advisory	0.4	0.3, 0.7			0.001
	Ever had an advisory compared to not having had an advisory for those who use a ground water source	7.3	4.6, 11.5			< 0.001
	Ever had an advisory compared to not having had an advisory for those who do not use a ground water source	3.0	1.6, 5.6			0.001
Variances of Random Effects		Variance	SE
Postal Code		0.039	0.080
Region		0.003	0.030
Number of observations = 1734

Note:

OR: odds ratio. CI: confidence interval. SE: standard error.

^a Type III test of fixed effect for interaction

Figure 3. Predicted probabilities of believing the tap water is not safe to drink for the interaction between reporting a drinking water advisory and use of a ground water source, separated by whether or not an aesthetic complaint about the water was identified, and averaged over the use of a community water supply. CI: confidence interval.

Grouping by postal code accounted for 1.0% of the total variance in the model for the perception that tap water is not safe, while region accounted for 0.1%. This represents an improvement over the null model (in which postal code accounted for 6.4% of the total variance, and region 0.9%), suggesting that the fixed effects in this model explain most of the differences in the perception that tap water is not safe to drink between postal codes and regions of the province.

Fear that water will become contaminated

After accounting for other risk factors, respondents who reported having had a drinking water advisory were more likely to fear that their household’s water would become contaminated compared to those who did not experience an advisory (OR = 2.2, 95% CI 1.7–2.8, p < 0.001). Those who used a community water supply were less likely to fear their drinking water would become contaminated than those who used another source (OR = 0.5, 95% CI 0.4–0.6, p < 0.001). Those who had lived in the community for more than 10 years were less likely to fear contamination compared to those who had lived in the community for 10 years or less (OR = 0.8, 95% CI 0.6–1.0, p = 0.04). Age was associated with the fear that water will become contaminated (p = 0.003). Compared to the over-65 age group, the 45–54 age group (OR = 1.7, 95% CI 1.3–2.3, p < 0.001) and the 55–64 age group (OR = 1.4, 95% CI 1.1-1.8, p = 0.02) were more likely to report fear that the tap water will become contaminated. Estimates in this model were adjusted for the presence of children in the home.

Perception that someone became ill from drinking the household’s tap water

Respondents who reported a drinking water advisory (OR = 6.0, 95% CI 3.2–11.4, p < 0.001) were more likely to report that someone had ever become ill from drinking the household tap water than those who did not, after accounting for other risk factors. A similar association was found for those who reported any aesthetic complaint about the tap water compared to those who did not (OR = 3.7, 95% CI 2.1–6.5, p < 0.001). Those who reported use of a community water supply were less likely to report that someone had become ill as a result of drinking the household’s tap water compared to those who did not use a community supply (OR = 0.2, 95% CI 0.1–0.4, p < 0.001). Those who reported using a private water supply were also less likely to believe that their tap water had made someone ill compared to those who did not (OR = 0.3, 95% CI 0.2–0.7, p = 0.005).

Summary of risk factors

A comparison of the risk factors included in each of the final models is presented in Table 6. Having had a water advisory was a risk factor in all of the models, and each model also included at least one variable related to the tap water source or supply. Having an aesthetic complaint was included in both models in which it was assessed as a risk factor.

Table 6. Summary of the analyses for all water-quality or risk-perception outcomes, including the number of observations used and the risk factors associated with the final model for each outcome (effect modifications italicized).

Model	n	Increased odds of having outcome	Decreased odds of having outcome
Any aesthetic complaint	1865	Ever had advisory	Trucked water delivery
			Male
			Increasing age
			Lived in community > 10 years
Tap water not safe to drink	1734	Aesthetic complaint	Community water supply
		Had a water advisory –magnitude of effect greater for ground water users	Use a ground water source, but only for those who never had water advisory
Fear water will become contaminated^a	1784	Ever had a water advisory	Community water supply
		Age 45–64 more likely than ≥ 65	Lived in community > 10 years
Anyone ever ill from tap water	1671	Aesthetic complaint	Private water supply
		Ever had an advisory	Community water supply

^a Adjusted for whether or not children reside in home.

Discussion

In this study, we examined different measures of perceptions related to water quality using quantitative techniques that allowed us to evaluate the combined effect of multiple risk factors and to consider the potential for similarities in opinions within communities. This study included several different regions in rural Saskatchewan, where residents use a variety of tap water supplies and sources, and where water supplies could be impacted by varied geology and land-use activities. Although the final models for each outcome differed, there were several themes that emerged. Each model included at least one risk factor related to the type of water supply used in the household, suggesting that some aspects of a household’s water supply are important in water quality and risk perception. Having had a drinking water advisory was also a common predictor of negative perceptions about water quality and related risks, as was having an aesthetic complaint in the models where it was evaluated. Age was identified as a risk factor for having any aesthetic complaint about the tap water as well as fearing that tap water will become contaminated, and gender was a risk factor for having any aesthetic complaint.

Aesthetic complaints

The aesthetic qualities of water can influence our perceptions of water quality (Doria et al. 2009; Doria 2010), as well as water consumption patterns (Levallois et al. 1999; Dupont et al. 2010; Saylor et al. 2011). We investigated potential risk factors associated with having reported any aesthetic complaint about tap water. Having had any type of water advisory, the length of time in the community and using a trucked water supply were important factors in reporting any aesthetic complaint, along with age and gender.

In the present study, 25.3% of respondents indicated some aesthetic complaint about their tap water, be it unpleasant taste, odor, discoloration or cloudiness. Previous studies have examined distinct aesthetic qualities (e.g. taste and odor separately); however, because we were interested in the presence of any aesthetic complaint, we grouped our responses about aesthetic qualities into one index variable for analysis. Just 12.5% of our respondents were dissatisfied with the odor of their tap water and 14.5% were dissatisfied with its taste, compared to a recent cross-Canada study in which 33% of participants were dissatisfied with the odor of their tap water, and 31% were dissatisfied with the taste (Dupont et al. 2010). The frequency of aesthetic complaints was similar in both ground water and surface water users.

Of the variables related to tap water supply, only the use of a trucked water supply was significant in the final model for having an aesthetic complaint, and it decreased the likelihood of having any aesthetic complaint. Possible explanations for this effect are not clear, and could be investigated further.

To our knowledge, this is the first study to examine the potential impact of experiencing water advisories on subsequent perceptions of water quality and safety. In our analysis, having had a water advisory increased the likelihood having an aesthetic complaint about the tap water. However, the mechanism behind this association is not clear. Objectionable aesthetic qualities of water could be associated with events that lead to water advisories; for example, increased turbidity is a common trigger for advisories. However, because prior experience is an important influence on perceptions of quality and safety (Doria 2010), it could be that experiencing an advisory might heighten concern about water quality, and increase sensitivity to water’s aesthetic characteristics.

Living in an area for a longer duration reduced the likelihood of having an aesthetic complaint compared to shorter durations of residence. Familiarity has previously been reported to be an important factor in perception of water quality (Dietrich 2006; Doria 2010).

Females had higher odds of reporting dissatisfaction with the aesthetic qualities of tap water, which may be related to the tendency of women to attribute higher risks to hazards than males (Finucane et al. 2000). Respondents in the youngest age category also had greater odds of reporting dissatisfaction with the aesthetic qualities of their tap water. A previous study found that age was a significant risk factor in models of taste dissatisfaction, but the direction of the effect differed by community (Turgeon et al. 2004), suggesting that the relationships between age and water quality perceptions are complex and likely influenced by other factors.

Although there was not much variance attributable to differences between postal codes, and even less attributable to region, the included risk factors did not explain much of the variation in postal code and very little of the variation in region compared to the null model. This suggests that there are other unmeasured risk factors which might explain geographical differences in the perception of aesthetic qualities of drinking water in Saskatchewan which could be examined in future studies.

Perception that tap water is not safe

Nearly 12% of our respondents reported that they believed their tap water is not safe to drink, comparable to a recent US study in which 15% of respondents felt their tap water was unsafe to drink (Hu et al. 2011). Having any aesthetic complaint increased the likelihood of perceiving that tap water is not safe to drink, similar to findings in other studies (Jardine et al. 1999; Jones et al. 2005, 2007; Doria et al. 2009). While it has been recognized that taste and odor should not be ignored as indicators of water quality (Jardine et al. 1999), others have pointed out that aesthetic qualities are poor indicators of safety for a variety of reasons (Turgeon et al. 2004). For example, chlorine taste has been the source of taste dissatisfaction in other studies and has been associated with risk perception in those studies (Turgeon et al. 2004; Doria et al. 2009), despite the fact that chlorine should improve the microbiological safety of water. However, in the present study, respondents who used community treated water supplies were less likely to perceive health risks compared to those who did not use community supplies.

Having had a water advisory was identified as a risk factor for having an aesthetic complaint, and both were included as risk factors in the model for having an aesthetic complaint despite concerns of endogeneity among the risk factors. It is possible that each risk factor independently affects the perception of risk from the drinking water as well as having a confounding relationship. The perception of aesthetic qualities of drinking water has previously been identified as important in risk perception, and it is plausible that having had a water advisory would have a separate, and perhaps greater, impact on risk perception than on the presence of aesthetic complaints. This study was intended to identify important risk factors in risk perception related to drinking water among rural Saskatchewan residents, rather than estimating the exact magnitude of the relationships between the risk factors, warranting consideration of all potentially relevant risk factors.

The effect modification between use of a ground water source and having had a water advisory hints at the complexity of the relationships among different factors that affect perceptions of water quality and risk, and could reflect an understanding of the potential for long-term contamination of ground water sources compared to surface water.

Fear that water will become contaminated

Although relatively few respondents felt that their water was unsafe, a larger proportion (35.5%) reported fearing that their water will become contaminated in the future. Jones et al. (2006) reported that 41% of private water supply users in a southern Ontario community shared similar concerns. It is unclear from our results whether this is a general fear, or one based on the perception that there is an immediate risk to their water supply. Having experienced a water advisory was also important, lending support to the theory that it is an important risk factor in many aspects of water quality and risk perception. Use of a community water supply decreased the likelihood of fearing contamination of the water supply, as did living in an area for greater than 10 years, which perhaps are related to the issues of trust and familiarity in risk perception (Doria 2010). Being in the oldest age category also decreased the likelihood of reporting this concern compared to the intermediate age groups, but not when compared to the youngest age category.

Illness due to drinking tap water

No attempt was made to clarify the type of illness attributed to drinking the household’s tap water, nor whether drinking water had been confirmed as the cause of any illness; therefore, this outcome was considered a measure of perception of health risk from drinking tap water. Just over 3% of respondents believed that their household tap water had ever made anyone ill, similar to other results in Ontario (Jones et al. 2006) and Pennsylvania (Merkel et al. 2012). Having experienced a water advisory and having an aesthetic complaint were strongly associated with the perception that the tap water had been the source of illness, supporting the idea that prior experience with the water is an important factor in perception of risks related to drinking water (Doria 2010).

Because the use of a community water supply and a private water supply were not mutually exclusive or the only water supply options available, both types of supply were evaluated separately as risk factors. The inclusion of both in the final model for the perception that drinking the tap water had made someone ill suggests that the use of these types of water supply was associated with more confidence in the safety of the household tap water compared to other supply types (for example, public water stations or water delivered by truck).

Limitations

The choice to target specific regions of the province and the limited response rate to the questionnaire could have introduced selection bias and, therefore, these results might not be generalizable across rural Saskatchewan. While respondents did include residents of up to 24% of the rural municipalities in the province, females and the over-55 age groups were over-represented in the study population compared to the population in the areas to which the survey was targeted, suggesting that the generalizability of results across the population may be limited. However, the proportion of respondents who reported aesthetic complaints and fear of water contamination is comparable to other surveys done in Canada. It unclear if our study population disproportionately represents residents with awareness of issues regarding drinking water; however, the anonymous nature of the survey made more intensive evaluation of the characteristics of nonresponders impossible.

Our questionnaire did not include questions regarding education level or income, which limited our ability to assess the generalizability of our results. These could also potentially be important factors in the perceptions of water quality, although the importance and effect of these measures has varied in past studies (Turgeon et al. 2004; McSpirit and Reid 2011).

Given the complexity of the factors that influence water quality and risk perception, varied interpretation of the questions by different respondents could have introduced the potential for nondifferential misclassification, and limited our capacity to identify all of the potential risk factors. Also, many of the questions that directly queried aspects of quality and risk did not specify a time frame over which respondents should report their perceptions or concerns, so it was not always clear if respondents were indicating current, recent, past or average perceptions.

Conclusions

Qualitative reports have provided a basis for understanding quality and risk perception, but few other studies have attempted to quantify the risk factors associated with these perceptions. In this study, we used multivariable analysis to estimate the effects of individual risk factors as well as to evaluate differences among communities and geographic regions in the province. This study underscores the complexity of the numerous factors involved in the formation of perceptions of water quality and safety. Having experienced a water advisory was an important risk factor in all of our models of perceptions of quality and risk, providing evidence that this personal experience can impact trust in a water source. The aesthetic qualities of water were important risk factors in the models for perceived health risks. Some aspect of the type of household tap water supply was also included in each of the models, although the specific risk factor included, and its effects, were varied. Personal characteristics also had some influence on water quality and risk perception. Future work should seek to better understand the factors influencing water quality and risk perceptions, and how to address these factors to effectively implement public health education programs in rural areas about appropriate water management practices, the safety of rural water supplies and the need for regular water testing.

Acknowledgements

This research was made possible by the support of the Saskatchewan Health Research Foundation and the Interprovincial Graduate Student Fellowship program.