Primary Drinking Water Source and Acute Gastrointestinal Illness: New Mexico, 2007 Shawna J. Feinman, P. Barry Ryan, Barbara Toth, Wayne A. Honey & Julia W. Gargano Water Quality, Exposure and Health ISSN 1876-1658 Water Qual Expo Health DOI 10.1007/s12403-014-0148-0 1 23 Your article is protected by copyright and all rights are held exclusively by Springer Science+Business Media Dordrecht 2014 (outside the USA). This e-offprint is for personal use only and shall not be selfarchived in electronic repositories. If you wish to self-archive your article, please use the accepted manuscript version for posting on your own website. You may further deposit the accepted manuscript version in any repository, provided it is only made publicly available 12 months after official publication or later and provided acknowledgement is given to the original source of publication and a link is inserted to the published article on Springer's website. The link must be accompanied by the following text: "The final publication is available at link.springer.com”. 1 23 Author's personal copy Water Qual Expo Health DOI 10.1007/s12403-014-0148-0 ORIGINAL PAPER Primary Drinking Water Source and Acute Gastrointestinal Illness: New Mexico, 2007 Shawna J. Feinman · P. Barry Ryan · Barbara Toth · Wayne A. Honey · Julia W. Gargano Received: 3 June 2014 / Revised: 3 October 2014 / Accepted: 6 October 2014 © Springer Science+Business Media Dordrecht (outside the USA) 2014 Abstract The objectives of this study are to characterize New Mexico residents’ primary drinking water sources, consumption, and filter use by demographic characteristics, and to compare the 30-day prevalence of self-reported acute gastrointestinal illness (AGI) by water sources. We analyzed data on 6,600 adults surveyed in the 2007 New Mexico Behavioral Risk Factor Surveillance System. We estimated population frequencies and evaluated associations using chi-square tests and weighted multivariable logistic regression modeling. Over half (55 %) of individuals used public water as their primary drinking water source, 18 % used private wells, and 27 % used bottled water. Overall, 43 % of residents said they filtered their home tap water, which did not differ significantly by source. Compared to public water users, private well users had key demographic differences, including age, marital status, race, and education. The overall 30-day prevalence of AGI was 15 %. In models adjusted for demographic characteristics and health status indicators, individuals using well water had a non-significantly decreased odds of reporting AGI and seeking medical attention for AGI (odds ratio (OR) 0.83, 95 % CI 0.65–1.06 and OR 0.85, 95 % CI 0.41– 1.80). This baseline measure suggests private well users are not uniformly distributed throughout the New Mexico popS. J. Feinman · P. B. Ryan Rollins School of Public Health, Emory University, Atlanta, GA, USA S. J. Feinman · J. W. Gargano (B) Waterborne Disease Prevention Branch, Centers for Disease Control and Prevention, 1600 Clifton Rd. NE, MS C-09, Atlanta, GA 30329, USA e-mail: igc5@cdc.gov B. Toth · W. A. Honey Epidemiology & Response Division, New Mexico Department of Health, Santa Fe, NM, USA ulation. This information is useful when planning educational outreach to targeted populations. Our cross-sectional analyses did not reveal significant associations between primary drinking water source and AGI. Future epidemiologic studies including children and measuring the duration of exposure and water quality are needed to fully understand the health impacts of drinking untreated or undertreated water. Keywords Groundwater · Drinking water · Behavioral surveillance · Acute gastrointestinal illness · Surveys Introduction American Housing Surveys from 1997 to 2007 indicate 12.9 million to 15.6 million households used private well water as their primary source of drinking water (U.S. Census Bureau 2000, 2008). While the Safe Drinking Water Act (SDWA) empowers the Environmental Protection Agency (EPA) to regulate public water supplies, private water sources are not regulated under federal law. Instead, the responsibility to maintain and periodically test private well water quality falls upon well owners, who are not legally obligated by any federal regulations to perform such tests. Although no national data are available, a few studies have indicated that many well owners do not properly test or maintain their wells (Jones et al. 2005; Laflamme and Van Derslice 2004). The large number of individuals affected emphasizes the importance of understanding health risks associated with drinking from such sources and providing a more current assessment of the population at risk. Groundwater supplies, including those that supply private wells, are susceptible to pathogen contamination. A US Geological Survey study, which was conducted across the 123 Author's personal copy S. J. Feinman et al. US from 1991 to 2004, examined contamination of domestic well raw source water samples (i.e., sampled from the point of extraction). The study found that coliform bacteria were present in 34 % and E. coli were present in 8 % of sampled wells (DeSimone 2008). Multiple studies have found viral (Borchardt et al. 2003, 2012), bacterial (Zimmerman et al. 2001), and other indicators of fecal contamination (Strauss et al. 2001) in private water sources. Contaminated water supplies can result in illness when individuals ingest pathogens that might be in the water. Waterborne disease outbreak surveillance conducted in the United States between the years of 1971–2006 indicated that 31 % of drinking water-related outbreaks occurred as a result of using contaminated untreated groundwater. Among outbreaks involving public water systems, 26 % were due to using untreated or improperly treated groundwater that was contaminated at the source. Among outbreaks involving individual water systems (primarily private wells), the most common deficiency (83 %) was untreated or improperly treated contaminated groundwater (Craun et al. 2010). The millions of U.S. residents who depend on private wells might periodically or routinely be exposed to waterborne pathogens while assuming their water is safe (Jones et al. 2005). In addition, certain populations, such as the elderly and young children, are more susceptible to getting ill and therefore require special precautions (Committee on Envrionmental Health and Committee on Infectious Diseases 2009; U.S. Environmental Protection Agency 2003). Promoting private well testing and maintenance is a worthwhile public health goal, but few studies have published populationbased information on the demographic characteristics of private well users to inform health promotion efforts. Such information would be especially important in a state such as New Mexico, where 20 % of the population receives their water from a private well and the population demographics are diverse (State of New Mexico 2014; U.S. Census Bureau 2012). This information could be useful in planning for specific public health interventions and increasing awareness of well maintenance guidelines. The objectives of our analysis were to characterize New Mexico residents’ primary drinking water sources by demographic characteristics, and to compare the 30-day prevalence of self-reported acute gastrointestinal illness (AGI) by residents’ primary drinking water sources. Methods We used data from the New Mexico 2007 Behavioral Risk Factor Surveillance System (BRFSS) survey regarding water consumption, AGI episodes, health status indicators, and demographic characteristics. A detailed description of the 123 BRFSS survey design and random sampling procedures is available elsewhere (Centers For Disease Control and Prevention 2008). All analyses used survey procedures in SAS version 9.3 (SAS Institute Inc., Cary, NC) to account for the BRFSS complex survey design. The analyses appropriately weighted survey responses to represent the noninstitutionalized (i.e. not residing in institutional group quarters or facilities such as correctional institutions, juvenile facilities, skilled nursing facilities, and other long-term care living arrangements) adult population of New Mexico (Behavioral Risk Factor Surveillance System 2006). Respondents were asked whether they had experienced diarrhea (defined as 3 or more loose stools in a 24-h period) in the 30 days prior to the interview, and if so, whether they had sought medical attention; positive responses to these questions were defined as “AGI” and “AGI requiring medical attention,” respectively. Respondents were asked three questions about water: (1) their primary source of drinking water (public, private, or commercially produced), (2) whether they filter their home drinking water in any way, and (3) how many cups of home tap water they drink per day, which included water used in making hot beverages or juice. Demographic characteristics and health status indicators were both analyzed with respect to primary drinking water source and reported AGI. Health indicators were selected to take into account the overall health of respondents and to identify any underlying health conditions that might affect health behaviors. Sample frequencies were tabulated, and population percentages and 95 % confidence intervals (CI) were estimated using PROC SURVEYFREQ. Differences in proportions were assessed using the Rao-Scott chisquare test (SAS Institute Inc. 2008). The association between AGI and primary water source was investigated using logistic regression models. To address the complex survey design, logistic regression was carried out using the SAS procedure SURVEYLOGISTIC. Water source was retained in all models because this was the primary exposure of interest. Separate models of demographic and health status indicators associated with AGI were developed using manual backward selection (P < 0.05 to remain in the model); the variables selected in these models were combined into one model, and non-significant variables were removed. In addition, confounding was assessed by evaluating whether each variable changed the odds ratio for water source by ≥10 % and any demographic characteristic or health status indicator that met these criteria were added back into the final model. Modeling procedures were repeated for the outcome AGI requiring medical attention. Statistical significance was based on a P value of <0.05 or a 95 % confidence interval around an odds ratio that did not include 1. Cells with a minimum denominator cell size less than 50 were suppressed. All percentages were Author's personal copy Primary Drinking Water Source and AGI weighted to reflect the New Mexico non-institutionalized population. Results Out of an estimated total population of 1.3 million, a total of 6,606 non-institutionalized New Mexico adults residing in private residences with a landline telephone responded to the survey after being randomly selected for inclusion. The survey, before applying population weights, captured 2,457 males and 4,149 females aged 18–99 years, with a median age of 53. All questions used in this analysis had response rates between 90–100 %. However, only individuals who provided information regarding their primary drinking water source were included in this analysis (n = 5,963). With the exception of sex, and presence of children in the home, all respondent demographic characteristics tested were significantly different among primary drinking water sources (Table 1). Homes with children had a marginally significant lower proportion reliant on private well water than homes without children (15.8 vs. 19.5 %, P = 0.05). An estimated 40.4 % (95 % CI 38.5–42.2) of all households included children and 35.4 % (95 % CI 31.1–39.7) of homes relying on private wells included children (not shown). Income was also analyzed with primary drinking water sources, though there was no significant finding. The 30-day prevalence of AGI was 15.0 % (95 % CI 13.8–16.3) (Table 2). Of the residents with AGI, 11.2 % sought medical attention for their diarrhea and about half of those individuals provided a stool sample to their health care provider (not shown). In bivariate analyses, AGI did not differ significantly by sex, education, income, or number of children in household. AGI was significantly associated with age and marital status. The age groups with the highest AGI 30-day prevalence were 25–44 and 45–64 years (17.4 and 16.4 %, respectively), and the age group with the lowest AGI 30-day prevalence was 18–24 years (8.3 %). Respondents who were divorced or separated reported the highest 30-day prevalence of AGI (20.2 %), and respondents who were never married reported the lowest 30-day prevalence of AGI (12.2 %). High blood pressure and previously having a heart attack were not significantly associated with AGI 30day prevalence; having arthritis and being a current smoker were significantly associated with AGI. Overall, nearly 43 % of residents filtered their water; residents who used private sources as their primary water source were the most likely to filter their tap water (45.9 %) and residents whose primary water source was from commercial sources were the least likely to filter their tap water (39.7 %), but these differences were not statistically significant (not shown). The prevalence of AGI was nonsignificantly lower for people using private well sources (12.9 %) than for people using public water (15.7 %) or commercial water (16.3 %) (Table 3). AGI prevalence was similar among those who did and did not filter their tap water. The unadjusted odds ratio of reporting an AGI episode was 0.80 (95 % CI 0.63–1.02) for those who primarily used private well water and 1.05 (95 % CI 0.82–1.34) for those who primarily use commercial water compared to those who primarily use public water sources (Table 4). In the fully adjusted model, age, arthritis, and smoking remained significantly associated with AGI. Residents aged 25–44 had a significantly higher odds of AGI than residents aged 18–24 [OR 2.03 (95 % CI 1.10–3.75)]. Persons with arthritis had an increased odds of AGI (OR 1.90, 95 % CI 1.55–2.33) compared to persons without arthritis. Current smokers had higher odds of AGI than non-smokers (OR 1.61, 95 % CI 1.28–2.03). Adjusting for these variables did not meaningfully alter the magnitude of the association between primary water sources and AGI. Water source was not significantly associated with seeking attention for AGI in unadjusted or adjusted models (Table 5). In the unadjusted model, the odds ratio for AGI episode was 0.74 (95 % CI 0.36–1.52) for those who primarily used private well water and 0.84 (95 % CI 0.49–1.44) for those who primarily use commercial water, compared to those who primarily used public water sources. After adjusting for household income, history of heart attack, arthritis, or hypertension, and age, the magnitudes of the odds ratios for primary water source were attenuated and remained non-significant. Discussion AGI prevalence did not vary significantly by primary water supply in this cross-sectional survey, although well-water users had non-significantly lower odds of AGI than did public water users. To our knowledge, this study is the first that uses BRFSS data to quantify AGI burden associated with private well use within a population. Anticipating that the lack of regulation on private wells would result in low frequency of water quality monitoring and poor maintenance of wells, we had hypothesized that AGI would be more common among private well users. This hypothesis was informed by results of surveys and focus groups exploring well-owners’ management tendencies (Jones et al. 2005; Laflamme and Van Derslice 2004), as well as studies that linked AGI outbreaks to contaminated groundwater sources (Brunkard et al. 2011; Craun et al. 2010; Fong et al. 2007; Gallay et al. 2006; Kvitsand and Fiksdal 2010; Richards 2005; Said et al. 2003; Schuster et al. 2005), detected pathogens or indicators of fecal contamination in well water used for drinking (Borchardt et al. 2003, 2012; Lambertini 123 Author's personal copy S. J. Feinman et al. Table 1 Percentage of residents relying on different primary sources of drinking water by demographic characteristics and primary health indicators, New Mexico, 2007 Variables Total P valueb Public water Private well water Commercial water Na Na % (95 % CI)a Na % (95 % CI)a 18.1 (16.7–19.4) 1,500 26.7 (25.1–28.4) – 0.19 % (95 % CI)a 3,321 55.2 (53.4–57.0) 1,142 Male 1,264 56.0 (53.1–58.9) 460 18.8 (16.5–21.1) Female 2,057 54.4 (52.2–56.7) 682 17.4 (15.7–19.02) Sex 475 25.2 (22.5–27.9) 1,025 28.2 (26.2–30.2) Age (in years) 18–24 119 45.5 (37.6–53.3) 37 18.1 (11.4–24.8) 101 36.5 (29.0–44.0) 25–44 842 55.4 (52.3–58.5) 258 14.8 (12.7–16.9) 472 29.7 (26.8–32.6) 45–64 1,362 54.6 (52.2–57.1) 540 20.7 (18.8–22.6) 649 24.7 (22.6–26.8) 998 63.8 (60.9–66.7) 307 19.0 (16.6–21.3) 278 17.2 (14.9–19.6) 1,893 57.1 (55.0–59.2) 712 18.8 (17.3–20.3) 817 24.1 (22.3–26.0) 594 55.5 (51.4–59.6) 179 15.3 (12.6–18.1) 306 29.2 (25.3–33.0) 65+ <0.0001 Marital status Married/living together Separated/divorced Widowed 466 62.0 (57.2–66.8) 123 15.4 (12.3–18.5) 147 22.6 (18.0–27.2) Never married 362 46.6 (40.8–52.4) 123 18.0 (13.0–23.0) 225 35.5 (29.8–41.1) <High school diploma 375 52.7 (47.5–58.0) 114 12.7 (9.8–15.6) 264 34.6 (29.6–39.5) High school diploma/GED 887 52.8 (49.1–56.4) 330 20.0 (17.2–22.9) 403 27.2 (23.8–30.6) Some college 865 52.4 (48.8–56.1) 301 19.4 (16.2–22.5) 399 28.2 (24.9–31.5) 1,190 60.7 (57.9–63.5) 396 17.3 (15.4–19.3) 433 22.0 (19.5–24.4) <0.0001 Education College graduate <0.0001 Children in household None 1 or more 2265 54.3 (52.1–56.6) 837 19.5 (17.8–21.3) 963 26.1 (24.0–28.2) 1,056 56.5 (53.4–59.5) 305 15.8 (13.6–18.1) 537 27.7 (24.9–30.4) 0.05 Race/ethnicityc 176 46.6 (39.4–53.7) 83 18.5 (13.5–23.4) 160 35.0 (28.4–41.6) Hispanic American Indian or Alaska Native 1,026 54.9 (51.8–58.1) 310 14.2 (12.1–16.3) 572 30.8 (27.9–33.8) White 2,038 56.6 (54.3–58.9) 728 21.1 (19.1–23.1) 709 22.3 (20.2–24.3) Yes 1,095 57.3 (54.2–60.4) 373 19.5 (16.8–22.2) 448 23.2 (20.5–25.8) No 2,220 54.4 (52.2–56.6) 767 17.5 (15.9–19.2) 1,052 28.1 (26.0–30.1) Yes 202 66.8 (60.3–73.2) 52 15.8 (10.9–20.7) 55 17.4 (12.2–22.5) No 3,098 54.7 (52.9–56.6) 1,076 18.0 (16.6–19.5) 1,442 27.2 (25.5–29.0) Yes 1,226 59.3 (56.5–62.1) 397 18.8 (16.4–21.2) 486 21.9 (19.6–24.1) No 2,091 53.7 (51.4–56.0) 741 17.7 (16.0–19.4) 1,009 28.6 (26.4–30.7) Yes 645 57.7 (53.6–61.8) 212 16.1 (13.1–19.2) 298 26.2 (22.6–29.9) No 2,666 54.6 (52.6–56.6) 927 18.5 (17.0–20.1) 1,197 26.8 (25.0–28.7) <0.0001 High blood pressure 0.02 Heart attack <0.01 Arthritis <0.001 Current smoker 0.32 Percentages might not add up to 100 % due to rounding a N represents the sample frequencies. Row percentages and 95 % confidence intervals (CI) are weighted to account for sampling design and adjust the sample demographics to the age, gender, and region of residence of the adult population; see text for details b P value from Rao-Scott chi-square test. Null hypothesis is no association between primary water source and row variable c Race/ethnicities of Asian or Pacific Islander and Black or African American were removed from the analysis due to small sample size 123 Author's personal copy Primary Drinking Water Source and AGI Table 2 30-day prevalence of acute gastrointestinal illness by respondent demographic characterization and primary health status indicators, New Mexico, 2007 Variables Na % (95 % CI)a P valueb Total 993 15.0 (13.8–16.3) – Male 367 15.1 (13.1–17.2) 0.91 Female 626 15.0 (13.5–16.4) Sex Age (in years) 18–24 26 25–44 279 17.4 (15.0–19.8) 8.3 (4.0–12.7) 45–64 454 16.4 (14.6–18.1) 65+ 234 13.1 (11.2–15.1) Married/living together 542 15.1 (13.6–16.6) Separated/divorced 231 20.2 (16.8–23.5) Widowed 106 13.1 (10.1–16.1) Never married 110 12.2 (8.5–15.9) <0.001 Marital status <0.01 Education <High school diploma 103 12.9 (9.5–16.3) High school diploma/GED 254 14.2 (11.7–16.8) Some college 303 17.0 (14.4–19.6) College graduate 332 15.0 (13.1–16.9) 0 677 15.6 (14.0–17.1) 1 or more 316 14.3 (12.2–16.4) 77 13.9 (10.1–17.7) Hispanic 304 13.5 (11.5–15.6) White 591 16.3 (14.6–18.0) Yes 353 16.7 (14.4–19.1) No 636 14.4 (12.9–15.8) Yes 66 19.7 (14.1–25.4) No 916 14.8 (13.5–16.1) 0.23 Children in household 0.35 Race/ethnicityc American Indian or Alaska Native a N represents the sample frequencies, and percentages and 95 % confidence intervals (CI) are weighted to account for sampling design and adjust the sample demographics to the age, gender, and region of residence of the adult population; see text for details b P value from Rao-Scott chi-square test for differences in proportion with AGI across categories c Race/Ethnicities of Asian or Pacific Islander and Black or African American were removed from the analysis due to small sample size 0.03 High blood pressure 0.08 Heart attack 0.06 Arthritis Yes 455 20.9 (18.6–23.3) No 537 12.8 (11.3–14.3) <0.0001 Current smoker Yes 258 20.0 (16.9–23.2) No 731 13.7 (12.4–15.1) et al. 2012; Strauss et al. 2001; Zimmerman et al. 2001), and associated land parcels served by private wells with mapped cases of parasitic and bacterial gastrointestinal infections (Uhlmann et al. 2009). Although we considered primary water source, amount of water consumed, and water filtration habits, and evaluated two definitions of AGI, we did not identify any significant association between water source and AGI. Residents who habitually drink contaminated water might be less likely to <0.0001 develop symptoms of illness than visitors who consume the same water because frequent exposure to pathogens confers some level of immunity or reduced likelihood of symptomatic infections (Blumberg et al. 2011; Frost et al. 1998; Isaac-Renton et al. 1994; Macler and Merkle 2000; Simmons et al. 2013). The BRFSS survey did not include short-term visitors or children (who by definition have had a shorter duration of exposure to their home water supply), so our study population might not have included individuals most 123 Author's personal copy S. J. Feinman et al. Table 3 30-day prevalence of AGI according to primary drinking water source, filtration, and consumption, New Mexico, 2007 Variables Total Residents with AGI Na % (95 % CI)a Na AGI Prevalence % (95 % CI)a Public 3,321 55.2 (53.4–57.1) 540 15.7 (14.0–17.4) Well 1,142 18.1 (16.7–19.4) 176 12.9 (10.6–15.3) Commercial 1,500 26.7 (25.1–28.4) 246 16.3 (13.4–19.2) Yes 2,408 42.8 (41.0–44.6) 396 15.6 (13.7–17.5) No 3,553 57.2 (55.4–59.0) 569 15.4 (13.6–17.2) 0–4 cups/day 2,081 37.2 (35.4–39.0) 362 15.2 (13.3–17.2) 5+ cups/day 3,882 62.9 (61.0–64.6) 600 15.6 (13.9–17.2) P valueb Primary water source 0.19 Water filter 0.88 Water consumption 0.78 Filtered by source Public, filtered 1,401 24.0 (22.5–25.5) 229 15.6 (13.2–18.0) Public, unfiltered 1,905 31.4 (29.7–33.1) 311 15.8 (13.5–18.2) 8.3 (7.2–9.4) 71 12.7 (9.1–16.2) Well, filtered 465 Well, unfiltered 670 9.8 (8.7–10.8) 105 13.3 (10.2–16.4) Commercial, filtered 530 10.5 (9.4–11.7) 93 17.2 (12.7–21.8) Commercial, unfiltered 952 16.0 (14.6–17.4) 151 16.0 (12.2–19.8) 0.59 Water consumption by source Public, 0–4 cups 856 15.5 (14.2–16.9) 151 15.5 (12.4–18.6) Public 5+ cups 2,452 39.9 (38.0–41.5) 387 15.7 (13.7–17.8) Well, 0–4 cups 278 4.9 (4.1–5.7) 46 13.5 (8.9–18.2) Well 5+ cups 856 13.1 (11.9–14.3) 130 12.8 (12.0–15.5) Commercial, 0–4 cups 938 16.8 (15.4–18.2) 163 15.2 (12.6–18.3) Commercial 5+ cups 545 9.9 (8.6–11.0) 80 18.3 (12.6–24.0) 0.48 Water consumption by filtration 0–4 cups, filtered 747 14.6 (13.3–16.0) 125 15.1 (11.6–18.5) 1,318 22.6 (21.0–24.1) 236 15.4 (12.9–17.9) 5+ cups, filtered 1,653 28.3 (26.6–29.8) 270 15.9 (13.6–18.2) 5+ cups, unfiltered 2,205 34.6 (32.8–36.3) 329 15.4 (13.0–17.9) 0–4 cups, unfiltered 0.98 aN represents the sample frequencies, and percentages and 95 % confidence intervals (CI) are weighted to account for sampling design and adjust the sample demographics to the age, gender, and region of residence of the adult population; see text for details b P value from Rao-Scott chi-square test for differences in proportions with illness across exposure categories at risk of waterborne illnesses. No information was gathered on duration of residence, so it was not possible to test whether residents who moved into the area recently had a higher prevalence of illness than long-term residents. The age group most likely to report AGI was 25–44 year olds. This was somewhat surprising in light of research indicating that the very young and the very old are most susceptible to AGI (Committee on Envrionmental Health and Committee on Infectious Diseases 2009; Jones et al. 2007). Most demographic characteristics investigated differed significantly by primary drinking water source, indicating that primary drinking water sources are not uniformly distributed throughout the population. This might be explained by highly variable population density between different areas 123 of the state (i.e. those who dwell in the city are demographically different from those who reside in rural areas), and typically areas of greater population density use public water, while areas less densely populated use private well sources. The analysis of income indicated that income was not significantly related to AGI or source of drinking water. Demographic information should be considered when developing and distributing educational materials for well owners. For example, the survey found that 40 % of well-water users (not shown) either had only a high school diploma/GED or less than high school diploma, 69 % were between the ages of 25–64, and 65 % had no children living in the household. Therefore, educational materials should aim to reduce technical wording and highlight potential health risks for all ages, Author's personal copy Primary Drinking Water Source and AGI Table 4 Unadjusted and adjusted odds of reporting AGI by primary drinking water source, demographic characteristics, and health status indicators, New Mexico, 2007 Variables Unadjusted n = 5,937 Adjusted n = 5,906 OR OR 95 % CI 95 % CI Primary water source Public 1.00 1.00 Well 0.80 0.63 1.02 0.83 0.65 1.06 Commercial 1.05 0.82 1.34 1.11 0.86 1.42 Age (years) 18–24 1.00 25–44 2.03 1.10 3.75 45–64 1.62 0.89 2.95 65+ 1.16 0.62 2.16 1.55 2.33 1.28 2.03 Arthritis No 1.00 Yes 1.90 Current smoker Table 5 Unadjusted and adjusted model odds of seeking medical attention for AGI by primary drinking water source, demographic characteristics, and health status indicators, New Mexico, 2007 No 1.00 Yes 1.61 Variables Unadjusted n = 5,937 Adjusted n = 5,307 OR OR 95 % CI 95 % CI Primary water source Public 1.00 Well 0.74 0.36 1.52 0.85 1.00 0.41 1.80 Commercial 0.84 0.49 1.44 0.97 0.54 1.74 Less than $20,000 0.99 0.41 2.38 $20,000–$35,000 0.41 0.16 1.03 $35,000–$50,000 1.00 $50,000–$75,000 0.63 0.23 1.77 $75,000 or more 0.22 0.08 0.62 1.49 6.32 3.27 12.31 0.70 3.16 15.28 Household income Heart attack No 1.00 Yes 3.07 Arthritis No 1.00 Yes 6.34 Hypertension No 1.00 Yes 1.49 Age (years) 18–24 1.00 25–44 3.01 0.59 45–64 1.32 0.26 6.57 65+ 0.91 0.17 4.98 123 Author's personal copy S. J. Feinman et al. not just the very young or elderly. On the other hand, 35 % of households on private well water did have children living in the household. The American Academy of Pediatrics warns that children on private water sources, especially those that are not properly maintained or monitored, have an increased risk of developing AGI symptoms, and are more vulnerable than adults to some water contaminants (Committee on Envrionmental Health and Committee on Infectious Diseases 2009). Households with children and on private water sources might need additional, targeted educational information. In addition, discovering that some health indicators differ by primary water source does not necessarily suggest that the water source is responsible for the measured health problems. Rather, these findings might highlight the health and demographic differences between city and rural dwellers. Water filtration devices might provide a false sense of security in terms of the microbiological safety of water. In this survey, water filtration devices had no significant association with AGI prevalence. The survey prompted respondents to recall any filter type, including pitcher-type filters that likely impact taste and odor but not microbiological quality of the water and reverse osmosis systems that would be expected to remove many microbial contaminants. While it is not clear which residents were drinking water that was effectively filtering out pathogens, it is important to note that nearly half of respondents reported using some sort of filter at home for their home tap water. This presents another opportunity for consumer education, as many consumers might not be aware of the limitations of their home filtration systems. The 30-day prevalence of AGI in this study, 15 % (95 % CL = 13.8–16.3), was high compared to that found in other population surveys in North America. A similar survey conducted in British Columbia, Canada found a 30-day AGI incidence of 10 % (Jones et al. 2007). A report of the Foodborne Diseases Active Surveillance Network (FoodNet) described a 7.7 % average monthly prevalence of diarrhea from 1996 to 2003 among FoodNet states (Jones et al. 2006). To further explore this varied prevalence of diarrhea we conducted an additional regression analysis with the outcome restricted to cases who sought medical attention for their illness, and results with respect to water source were similar. In our multivariable models, multiple health status indicators were included in order to adjust for the confounding effects of individuals’ overall health on the association between water source and AGI. These analyses revealed the incidental finding that both arthritis and smoking were significantly associated with reporting AGI and seeking medical attention, which might be the result of certain drugs prescribed for arthritis (Bhatt et al. 2008), differences in hand hygiene behavior, or the result of the stomach pH being more favorable for bacteria in smokers (Bagaitkar et al. 2008). The BRFSS survey was not created for the purpose of investigating the association between household primary 123 drinking water source and reported AGI; therefore, the questions were not designed to be analyzed in this context. The survey elicited data on primary source of drinking water, which is likely a subset of all the water people consume. For example, a person might use a public water supply at home, drink commercially bottled water at work, and visit people or businesses that use private well sources. In this instance, multiple exposure sources could complicate the ability to accurately determine associated risks. In addition, we did not have information on the source waters (i.e., groundwater or surface water) or treatments used (e.g., filtration, disinfection) by public water systems serving the respondents, and analyzing all public water systems as one category might have masked meaningful heterogeneity in water supplies. Most of New Mexico’s public water systems obtain water from groundwater sources (United States Environmental Protection Agency 2013). Although the BRFSS expanded to include cell phones in later years, the 2007 survey only included households with landlines, which excluded 17.2 % of the New Mexico population and might have resulted in sampling bias if this population differed significantly and in some consistent or systematic fashion from households with landlines. For example, the age of the population captured may have been skewed because recent trends indicate younger individuals are less likely to have household telephone lines (Blumberg et al. 2011). The self-reported exposure and outcome data could result in differential or non-differential misclassification, which could either increase or decrease the likelihood of finding an association between water source and AGI. In addition, the 2007 NM BRFSS did not gather any data that might have helped to identify or rule out possible foodborne, zoonotic, person-to-person, or recreational waterassociated transmission of AGI. We are not aware of any data that compares groundwater quality in New Mexico to that in other states, and results of similar studies in other geographic locations might identify associations that this study did not. Finally, cross-sectional surveys, such as the BRFSS survey, cannot be used to establish causation of illnesses by chronic exposures, such as primary home drinking water source, because the temporal relationship between exposure and outcome cannot be established from a one-time survey of the population (Gordis 2009). The BRFSS is a population-based survey producing results representative of the New Mexico adult population. Therefore, the study results should be considered representative of the non-institutionalized adult population of New Mexico and cannot be generalized to other areas of the United States. However, these results suggest that similar data collection and analyses in other states may be important to the evaluation of the relationship between prevalence of self-reported AGI and use of unregulated private wells as a primary source of drinking water. While this analysis neither supports nor refutes the hypothesis that 30-day prevalence of AGI might be associ- Author's personal copy Primary Drinking Water Source and AGI ated with primary drinking water source in New Mexico, the results do provide valuable information regarding susceptible persons and water filtration habits within a population. This information might serve as a baseline measure and to direct educational efforts to selected vulnerable populations. Future epidemiologic studies that include children and measure the duration of exposure and water quality are needed to understand the health impacts of drinking untreated or undertreated water. Acknowledgments The authors would like to thank Dr. Gordana Derado for methodological input and constructive comments on an earlier draft of the manuscript. The authors would also like to thank Sarah Collier and Heidi Krapfl for constructive comments on manuscript drafts. This publication was supported in part by Cooperative Agreements (CA) 5U38EH000949 and 5U38EH00097 from the Centers for Disease Control and Prevention (CDC); the New Mexico BRFSS “Drinking Water Module” was supported by CA 1U38EH000183 from CDC. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the CDC. The manuscript does not contain clinical studies or patient data. References Bagaitkar J, Demuth DR, Scott DA (2008) Tobacco use increases susceptibility to bacterial infection. Tob Induc Dis. doi:10.1186/ 1617-9625-4-12 Behavioral Risk Factor Surveillance System (2006) Behavioral Risk Factor Surveillance System Operational and User’s Guide Version 3.0. ftp://ftp.cdc.gov/pub/data/brfss/userguide.pdf. Accessed 01 Oct 2014 Bhatt DL, Scheiman J, Abraham NS, Antman EM, Chan FKL, Furberg CD, Johnson DA, Mahaffey KW, Quigley EM (2008) ACCF/ACG/AHA 2008 Expert consensus document on reducing the gastrointestinal risks of antiplatelet therapy and NSAID use: a report of the american college of cardiology foundation task force on clinical expert consensus documents. J Am Heart Assoc. doi:10. 1161/CIRCULATIONAHA.108.191087 Blumberg SJ, Luke JV, Ganesh N, Davern ME, Boudreaux MH, Soderberg K (2011) Wireless substitution: state-level estimates from the National Health Interview Survey, January 2007–June 2010. Natl Health Stat Rep 39:1–26 28 Borchardt MA, Bertz PD, Spencer SK, Battigelli DA (2003) Incidence of enteric viruses in groundwater from household wells in Wisconsin. Appl Environ Microbiol 69:1172–1180. doi:10.1128/AEM. 69.2.1172 Borchardt MA, Spencer SK, Kieke BA Jr, Lambertini E, Loge FJ (2012) Viruses in nondisinfected drinking water from municipal wells and community incidence of acute gastrointestinal illness. Environ Health Perspect 120:1272–1279. doi:10.1289/ehp.1104499 Brunkard JM, Ailes E, Roberts VA, Hill V, Hilborn ED, Craun GF, Rajasingham A, Kahler A, Garrison L, Hicks L, Carpenter J, Wade TJ, Beach MJ, Yoder J (2011) Surveillance for waterborne disease outbreaks associated with drinking water—United States, 2007– 2008. MMWR Surveill Summ 60(12):38–68 Centers For Disease Control and Prevention (2008) About the BRFSS. http://www.cdc.gov/brfss/about.htm. Accessed 30 Dec 2012 Committee on Envrionmental Health and Committee on Infectious Diseases (2009) Drinking water from private wells and risks to children. Pediatrics 123:1599–1605. doi:10.1542/peds.2009-0751 Craun GF, Brunkard JM, Yoder JS, Roberts VA, Carpenter J, Wade T, Calderon RL, Roberts JM, Beach MJ, Roy SL (2010) Causes of outbreaks associated with drinking water in the United States from 1971 to 2006. Clin Microbiol Rev 23:507–528. doi:10.1128/CMR. 00077-09 DeSimone LA (2008) Quality of water from domestic wells in principal aquifers of the United States, 1991–2004: Overview of major findings. U.S. Geol. Survey Sci Investigations http://pubs.usgs. gov/circ/circ1332/includes/circ1332.pdf. Accessed 01 Oct 2014 Fong TT, Mansfield LS, Wilson DL, Schwab DJ, Molloy SL, Rose JB (2007) Massive microbiological groundwater contamination associated with a waterborne outbreak in lake Erie, South Bass Island. Ohio Environ Health Perspect 115:856–864. doi:10.1289/ ehp.9430 Frost FJ, Calderon RL, Muller TB, Curry M, Rodman JS, Moss DM, De La Cruz AA (1998) A two-year follow-up survey of antibody to cr yptosporidium in Jackson County, Oregon following an outbreak of waterborne disease. Epidemiol Infect 121:213–217 Gallay A, De Valk H, Cournot M, Ladeul B, Hemery C, Castor C, Bon F, Mègraud F, Le Cann P, Desenclos JC (2006) A large multipathogen waterborne community outbreak linked to faecal contamination of a groundwater system, France, 2000. Clin Microbiol Infect 12(6):561–570. doi:10.1111/j.1469-0691.2006.01441.x Gordis L (2009) Epidemiology. Saunders, Philadelphia Isaac-Renton JL, Lewis LF, Ong CS, Nulsen MF (1994) A second community outbreak of waterborne giardiasis in Canada and serological investigation of patients. Trans R Soc Trop Med Hyg 88(4):395–399 Jones AQ, Dewey CE, Doré K, Majowicz SE, McEwan SA, WaltnerToews D, Henson SJ, Mathews E (2005) Public perception of drinking water from private water supplies: focus group analyses. BMC public health 5:129–141. doi:10.1186/1471-2458-5-129 Jones AQ, Majowicz SE, Edge VL, Thomas MK, MacDougall L, Fyfe M, Atashband S, Kovacs SJ (2007) Drinking water consumption patterns in British Columbia: an investigation of associations with demographic factors and acute gastrointestinal illness. Sci Total Environ 388:54–65. doi:10.1016/j.scitotenv.2007.08.028 Jones TF, McMillian MB, Scallan E, Frenzen PD, Cronquist AB, Thomas S, Angulo FJ (2006) A population-based estimate of the substantial burden of diarrhoeal disease in the United States; FoodNet, 1996–2003. Epidemiol Infect 135:293–301 Kvitsand HML, Fiksdal L (2010) Waterborne disease in Norway: emphasizing outbreaks in groundwater systems. Water Sci Technol 61:563–571. doi:10.2166/wst.2010.863 Laflamme DM, Van Derslice JA (2004) Using the Behavioral Risk Factor Surveillance System (BRFSS) for exposure tracking: experiences from Washington state. Environ Health Perspect 112:1428– 1433 Lambertini E, Borchardt MA, Kieke BA Jr, Spencer SK, Loge FJ (2012) Risk of viral acute gastrointestinal illness from nondisinfected drinking water distribution systems. Environ Sci Technol 46(17):9299–9307. doi:10.1021/es3015925 Macler BA, Merkle JC (2000) Current knowledge on groundwater microbial pathogens and their control. Hydrogeol J 8:29–40 Richards A (2005) The Walkerton health study. Can Nurse 101(5):16– 21 Said B, Wright F, Nichols GL, Reacher M, Rutter M (2003) Outbreaks of infectious disease associated with private drinking water supplies in England and Wales 1970–2000. Epidemiol Infect 130(3):469– 479 SAS Institute Inc. (2008) SAS/STAT ® 9.2 User’s Guide. In SAS Institute Inc., Cary Schuster CJ, Ellis AG, Robertson WJ, Charron DF, Aramini JJ, Marshall BJ, Medeiros DT (2005) Infectious disease outbreaks related to drinking water in Canada, 1974–2001. Can J Public Health 96(4):254–258 123 Author's personal copy S. J. Feinman et al. Simmons K, Gambhir M, Leon J, Lopman B (2013) Duration of immunity to norovirus gastroenteritis. Gastroenteritis Emerg Infect Dis 19(8):1260–1267. doi:10.3201/eid1908.130472 State of New Mexico (2014) NM EPHT environmental exposure: drinking water quality. https://nmtracking.org/environ_exposure/ water-qual/. Accessed 24 Feb 2014 Strauss B, King W, Ley A, Hoey JR (2001) A prospective study of rural drinking water quality and acute gastrointestinal illness. BMC Public Health 1:8 U.S. Census Bureau (2000) American Housing Survey for the United States: 1997. http://www.census.gov/prod/99pubs/h150-97.pdf. Accessed 4 Mar 2014 U.S. Census Bureau (2008) American Housing Survey for the United States: 2007. http://www.census.gov/housing/ahs/files/ ahs07/h150-07.pdf. Accessed 4 Mar 2014 U.S. Census Bureau (2012) New Mexico. http://quickfacts.census.gov/ qfd/states/35000.html. Accessed 21 Sept 2012 U.S. Environmental Protection Agency (2003) Water on tap: what you need to know. http://water.epa.gov/drink/guide/upload/book_ waterontap_full.pdf. Accessed 4 Mar 2014 123 Uhlmann S, Galanis E, Takaro T, Mak S, Gustafson L, Embree G, Bellack N, Corbett K, Isaac-Renton J (2009) Where’s the pump? Associating sporadic enteric disease with drinking water using a geographic information system, in British Columbia, Canada, 1996–2005. J Water Health 7(4):692–698. doi:10.2166/wh.2009. 108 United States Environmental Protection Agency (2013) Safe drinking water information system. http://www.epa.gov/enviro/facts/ sdwis/search.html. Accessed 4 Mar 2014 Zimmerman TM, Zimmerman ML, Lindsey BD (2001) Relation between selected well-construction characteristics and occurrence of bacteria in private household-supply wells, south-central and southeastern Pennsylvania. U.S. Geol Survey http://www.wilkes.edu/include/WaterResearch/PDFs/3676/ Bacteria%20Well%20Water.pdf. Accessed 4 Mar 2014