- Email: [email protected]
Assessing the infection risk of Giardia and Cryptosporidium in public drinking water delivered by surface water systems in Sao Paulo State, Brazil
Science of the Total Environment 442 (2013) 389–396
Contents lists available at SciVerse ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Assessing the infection risk of Giardia and Cryptosporidium in public drinking water delivered by surface water systems in Sao Paulo State, Brazil Maria Ines Z. Sato a,⁎, Ana Tereza Galvani a, Jose Antonio Padula a, Adelaide Cassia Nardocci b, Marcelo de Souza Lauretto c, Maria Tereza Pepe Razzolini b, Elayse Maria Hachich a a b c
CETESB — Companhia Ambiental do Estado de Sao Paulo, Av. Prof. Frederico Hermann Jr., 345, São Paulo, SP 05459-900, Brazil Faculdade de Saude Publica, Universidade de Sao Paulo, Av. Dr Arnaldo 715 1o andar, Sao Paulo, SP 01246-904, Brazil EACH — Escola de Artes, Ciencias e Humanidade, Universidade de Sao Paulo, R. Arlindo Bettio, 1000, São Paulo, SP 03828-000, Brazil
H I G H L I G H T S ► ► ► ► ►
Giardia was detected in high densities in source waters of urbanized watersheds. Cryptosporidium was seldom found and the oocysts densities were usually low. Giardia and Cryptosporidium infection risk was higher than 1/10,000. Giardia risk infection was greater for adults than that observed for children. The implementation of Water Safe Plan is a tool needed for ensuring improved water.
a r t i c l e
i n f o
Article history: Received 17 July 2012 Received in revised form 28 September 2012 Accepted 30 September 2012 Available online 22 November 2012 Keywords: Risk assessment Giardia Cryptosporidium Drinking water Source water
a b s t r a c t A survey of Giardia and Cryptosporidium was conducted in surface water used as drinking water sources by public water systems in four densely urbanized regions of Sao Paulo State, Brazil. A Quantitative Microbial Risk Assessment, based on protozoa concentrations, was performed to estimate the probability of protozoa infection associated with drinking water ingestion. A total of 206 source water samples were analyzed over a 24 month period using the USEPA Method 1623. The risk of infection was estimated using an exponential dose response model, children and adults exposure and a gamma distribution for (oo)cyst concentrations with three scenarios for treating censored data. Giardia was detected in 102 of the samples, and 19 of them were also positive for Cryptosporidium, with maximum concentrations of 97.0 cysts/L and 6.0 oocysts/L, respectively. Risk distributions were similar for the three scenarios. In the four regions, the estimated risk of Giardia infection per year, for adults and children, ranged from 0.29% to 2.47% and from 0.08% to 0.70%, respectively. Cryptosporidium risk infection varied from 0.15% to 0.29% for adults and from 0.04% to 0.08% for children. In both cases, the calculated risk surpassed the risk of infection of 10 −4 (1:10,000) defined as tolerable by USEPA for a yearly exposure. The probability of Giardia infection was very close to the rates of acute diarrheic disease for adults (1% to 3%) but lower for children (2% to 7%). The daily consumption of drinking water was an important contributing factor for these differences. The Microbiological Risk Assessment carried out in this study provides an indication of infection risks by Giardia and Cryptosporidium in the population served by these source waters. Strategies for source water protection and performance targets for the water treatment should be established to achieve the required level of public health risk. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Microbiologically unsafe drinking water supplies are still a public health problem all over the world but may be critical for developing ⁎ Corresponding author at: CETESB — Companhia Ambiental do Estado de Sao Paulo, Departamento de Analises Ambientais, Av. Prof. Frederico Hermann Jr., 345, Alto de Pinheiros, Sao Paulo, SP 05459-900, Brazil. Tel.: +55 11 3133 3541; fax: +55 11 3133 3982. E-mail addresses: [email protected], [email protected] (M.I.Z. Sato), [email protected] (A.T. Galvani), [email protected] (J.A. Padula), [email protected] (A.C. Nardocci), [email protected] (M.S. Lauretto), [email protected] (M.T.P. Razzolini), [email protected] (E.M. Hachich). 0048-9697/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2012.09.077
countries, especially in big cities, where the lack of adequate health services, sanitation and safe water combined with high population density and poor quality housing prevail in the peri-urban areas. Many metropolitan regions of Sao Paulo State in Brazil, the richest State of the country, share these features. According to health surveillance data, acute diarrheal diseases incidence is 1.2% in the population and reach 5% in children under 5 years old (CVE, 2010). Drinking water supplies for Sao Paulo State population come predominantly from surface water sources and some of these water bodies are heavily contaminated by discharge of raw or partially treated sewage from nearby towns.
390
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
Among the waterborne pathogens, protozoa pose major challenges to design and maintenance of safe water supplies (Zmirou-Navier et al., 2006). The protozoan parasites Giardia lamblia and Cryptosporidium hominis are recognized as important waterborne disease pathogens and are associated with severe gastrointestinal illness. Craun et al. (2006) reported that these parasites have been the most frequently identified etiologic agents in the last 12 years in the United States. A recent worldwide review of waterborne outbreaks caused by protozoan parasites (Karanis et al., 2007) provides detailed information regarding date and location, estimated number of cases and suspected causes of the outbreaks. According to this survey, out of 325 outbreaks, 32% (104) were associated with drinking water systems contaminated with G. lamblia, and 23.7% (77) had Cryptosporidium parvum or Cryptosporidium sp. as the etiologic agent. Deficiencies in the water treatment processes were the most cited reasons. Such deficiencies included gaps in the protective barriers and poorly operated treatment and disinfection systems. In Brazil, cryptosporidiosis and giardiasis represent an important cause of morbidity in children from 0 to 5 years (Carvalho-Almeida et al., 2006; Franco and Cordeiro, 1996; Gonçalves et al., 2006). Teixeira et al. (2007) reported a prevalence of Giardia in 18% of children in a sub-standard settlement in Juiz de Fora city (Brazil). Prado et al. (2003) in a study about risk factors for Giardia infection in pre-school children (aged 2–45 months) in the city of Salvador (Brazil) found that 13.7% (95/694) of them were infected with Giardia duodenalis. According to Mascarini and Donalisio (2006) there is high prevalence of giardiasis among children who are
attended in daycare centers in Brazil, mainly in those ones located in poor areas. The objectives of this study were to survey the prevalence of the protozoa Giardia and Cryptosporidium in surface water used as drinking water sources by public water systems in four densely populated regions of Sao Paulo State and to assess the risk of protozoan infection for the population supplied with drinking water from these sources. 2. Material and methods 2.1. Sampling The water sampling sites were situated in nine different watersheds at eastern region of the State covering the Metropolitan Region of Sao Paulo (MRSP), Metropolitan Region of Campinas (MRC), Metropolitan Region of South Coast (MRSC) and Vale do Paraiba region (VPR), which together comprise 58% of the Sao Paulo state and 13% of the Brazilian population (Fig. 1). The main characteristics of each region selected are summarized in Table 1. Ten liter volumes of water were collected for Giardia and Cryptosporidium analysis during 24 months (bimonthly) in 28 collection sites located at the source water intake area of water treatment plants (WTP) being five sites at VPR, nine sites at MRC, eleven sites at MRSP and three sites at MRSC. The water treatment processes in all WTPs consisted of coagulation, flocculation, sedimentation, sand filtration, chlorination and fluoridation. The water samples were
Fig. 1. Location of the four regions evaluated in Sao Paulo State, southeast of Brazil.
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396 Table 1 Main characteristics of the study area. VPR Populationa Proportion of state population (%)a Population density (inhab./km2)a Geometric rate of annual population growth (%) a Urbanization rate (%)a Yearly infant mortality rate (per 1000 birth)a Gross Domestic Product (GDP) per capita (USD)a Contribution to state GDP (%)a Drinking water coverage (%)a Sewage collection (%)a Sewage treatment (%)a Acute diarrheic disease rate (b5 years old) (%)b Acute diarrheic disease rate total population (%)b
MRC
MRSP
MRSC
2,014,219 2,847,077 19,847,879 1,682,435 5 7 48 4 141.95 0.83
780.95 1.82
2498.53 0.97
694.42 1.2
76.65 17.32
97.43 10.32
98.77 12.35
99.79 18.80
14,600
14,000
15,000
12,000
5 93.4 83 48 7
8 97.1 82 46 3
57 97.5 83 44 6
4 96.0 67 9 4
2
1
1
2
a Fundação SEADE (2011) — State Foundation of Data Analysis System (http:// www.seade.gov.br/). b CVE (2010) — Acute diarrheic disease rate was calculated from the number of the cases divided by correspondent aged population.
collected according to APHA (1998), kept on ice for transportation and processed within 24 h. 2.2. Protozoan analyses Protozoan parasites were detected using immunomagnetic separation–immunofluorescence assay (IMS-IFA) following USEPA 1623 Method (USEPA, 1999). Water samples were filtered through a 1 μm polycarbonate membrane (Whatman Inc.) and immunomagnetic separation was performed with Dynal reagents and equipment. The dissociation of the complex beads, cysts and oocysts was made by heat dissociation (Ware et al., 2003). The slides were stained using fluorescein conjugated monoclonal antibodies for Giardia and Cryptosporidium (Waterborne) and DAPI (4′, 6-diamidino-2-phenyl-indole, Sigma). The cysts and oocysts were identified and counted by immunofluorescence reaction and confirmed by DAPI fluorescence and DIC (differential interference contrast microscopy) using a Leica DMLD epifluorescence microscope equipped with bright field, phase contrast DIC and epifluorescence optics. Negative and positive control slides were also prepared. Recovery efficiency of the method was determined by spiking 10 L of purified water samples with EasySeed, according to the instructions of the manufacturer (BTFbio). The theoretical detection limit was calculated based on the assumption that at least one cyst or one oocyst could be detected by the microscopic examination of the slides, the entire packed pellet was examined and a 10 L volume of water sample was concentrated; therefore 1 oo(cyst) in 10 L or 0.1 oo(cyst) per liter. 2.3. Risk assessment procedure For conducting the risk assessment, raw water concentrations of Giardia and Cryptosporidium, log10 removal of (oo)cysts of Giardia and Cryptosporidium through the conventional water treatment, the dose–response model and the ingestion rate of water by children (b5 years old) and adults (>21 years old) were taken into account. 2.3.1. Concentrations of Giardia and Cryptosporidium in raw water In each region, concentrations of both organisms were adjusted via standard distributions. According to a preliminary analysis, lognormal and gamma distributions were considered and fitted by the maximum likelihood method. A graphical analysis and a comparison among maximum likelihood values suggested that the gamma distribution achieved
391
a better fitting, being therefore the distribution of choice for the risk analysis. A characteristic of the datasets was the occurrence of several left-censored observations — samples with concentrations below the theoretical detection limit (DL). Due to this reason, three scenarios were employed for treating censored data, as suggested by Clarke (1998) and Govaerts et al. (2005). In the first scenario, datasets were adjusted via maximum likelihood with left-censoring (Meeker and Escobar, 1998; Govaerts et al., 2005), implemented in R Package “fitdistrplus” (Delignette-Muller et al., 2010). Regions with less than five positive samples were not considered for this scenario. For the second scenario, the DL was assumed for censored data (b0.1(oo)cysts/L) whereas for the third scenario half of DL was used. For scenarios 2 and 3, filled-in data were adjusted via traditional (non-censored) gamma distributions. 2.3.2. Log 10 removal by the treatment process The removal efficiency of protozoan by the conventional treatment processes was assumed to be 3.0 log (99.9%) and 2.0 log (99%) for Giardia and Cryptosporidium, respectively as demonstrated by Nieminski (1997). All (oo)cysts were considered viable and infectious. 2.3.3. Dose response model The exponential dose–response model was used for both pathogens. The dose response parameter value (r-organism-specific infectivity parameter) adopted for Giardia was 0.01982 with a 95% Confidence Interval (CI) of 0.009798–0.03582 (Rose et al., 1991). For Cryptosporidium, r value was 0.00467 (95% CI: 0.00097–0.00915) (Haas et al., 1996). 2.3.4. Ingestion rate The water ingestion rates for the southeast region Brazilian population of different age categories (b5 and >21 years old) were calculated based on a research carried out by Kahn and Stralka (2009). A lognormal distribution for water ingestion rate with mean and standard deviation (SD) of 0.44 L/day (SD 0.92 L/day) for children and of 1.5 L/day (SD 0.80 L/day) for adults was assumed. 2.3.5. Risk calculation The daily risk of infection (Pi) was calculated by the equation: Pi ¼ 1− expð−r C IR TRÞ where: r C IR TR
dose–response model parameter; average concentration of Giardia or Cryptosporidium ((oo) cysts/L); ingestion rate (L/day); Log 10 treatment removal (3 logs for Giardia and 2 logs for Cryptosporidium).
Monte Carlo simulation was performed using R (R Development Core Team, 2011), providing 10,000 simulated values of Pi varying r, C and IR. Annual risk of infection (Pa) was computed by drawing 365 values Pi (1), Pi (2),…, Pi (365) from the 10,000 values of Pi, without replacement, and by applying the equation: 365 ðj Þ Pa ¼ 1−∏j¼1 1−Pi : This procedure was repeated 10,000 times to provide the annual risk simulated distribution Pa and the corresponding values of empirical mean, median and 95% probability interval (2.5% and 97.5% quantiles). 2.3.6. Sensitivity analysis In order to estimate the influence of the assumptions r, C and IR on risk results a sensitivity analysis was performed according to
392
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
the method described in Decisioneering (2001) and references therein. First, the rank correlation coefficients between every assumption and every estimated daily risk were computed. Then, the contributions to the variance were calculated by squaring the rank correlation coefficients for the assumptions and normalized to 100%. The contribution to the variance is an approximated method to estimate the percentage of the variance in the risk due to each assumption. 3. Results 3.1. Protozoa detection A total number of 206 samples collected throughout the 24 months, from 28 locations, were tested for both protozoa. Giardia was detected in 49.5% of the samples, in concentrations ranging from 0.1cyst/L to 97 cysts/L. Cryptosporidium was detected in 9.2% of the samples, in concentrations between 0.1 oocyst/L and 6 oocysts/L. Table 2 shows the number and percentage of samples according to the different protozoa concentration ranges. For Giardia most of the positive results were in the range of 0.1 to 10.0 cysts per liter, while for Cryptosporidium these values were in the range of 0.1–1.0 oocysts per liter. Although VPR and MRC regions have presented the highest percentage of positive samples for Giardia, the maximum concentration of this protozoan was detected in the Metropolitan Region of Sao Paulo (MRSP): densities of 97 cysts/L were found in samples collected in two consecutive months from the same waterbody. The highest frequency of positive samples and the highest concentrations of Cryptosporidium oocysts were observed in the Metropolitan Region of Campinas. This protozoan was not detected in any sample from VPR. Initial Precision and Recovery (IPR) performed for the Method 1623 showed recovery percentages of 39.4% for Giardia and of 41.3% for Cryptosporidium. These recovery rates were not used for the risk calculation. The adjusted gamma distribution concentration parameters (shape and rate) of each region and scenarios are shown in Table 3. Table 4 presents the estimated mean and standard deviation for the fitted distribution. Results indicated that, in each region, fitted distributions in the DL scenario have similar means for the three scenarios. In all regions, standard deviations for adjusted scenario were slightly higher than those in DL and half DL scenarios. 3.2. Risk assessment Tables 5 and 6 summarize the annual probability of infection by ingestion of cysts and oocysts for adults and children, computed in each region and scenario. Risk distributions were similar for the three scenarios and showed approximately a symmetric shape, as indicated by their means, medians and probability interval limits. All
Table 3 Parameters of gamma distribution concentrations per regions considering the three scenarios. Regions/scenarios
Giardia (cysts/L) All regions VPR MRC MRSP MRSC
Adjusteda Shape
Rate
0.1574 0.2473 0.1887 0.1058
0.0950 0.1789 0.0901 0.0595
c
c
Cryptosporidium (oocysts/L) All regions 0.0397 MRC 0.0758 c MRSP c MRSC a b c
DLb Shape
0.5485 0.5774 c c
Half DL Rate
Shape
Rate
0.4025 0.4917 0.4046 0.3576 0.9546
0.2369 0.3478 0.1900 0.1948 3.8182
0.3424 0.4198 0.3507 0.3020 0.6220
0.2045 0.3012 0.1664 0.1674 2.9618
1.5361 1.0386 1.6270 12.5429
9.6177 4.9799 10.3536 114.0318
0.8940 0.6706 0.9233 3.9024
7.8207 4.0138 8.3396 62.4335
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for this scenario.
values obtained were above the annual acceptable risk level of 10 −4 (0.01%) recommended by USEPA (1998). The median probability of Giardia infection for all regions was about 2% for adults and 0.6% for children, and the highest risk was estimated for the Metropolitan Region of Campinas, followed by Sao Paulo and Vale do Paraiba Regions. The probability of infection by Cryptosporidium was lower than that by Giardia. Due to the low frequency of positive samples for this parasite, the estimated risks showed the influence of the scenario choice, DL scenario presenting the more conservative results. The highest estimated risk was obtained in MRC region, followed by MRSP and MRSC regions. Tables 7 and 8 present the risk sensitivity indexes expressed as contribution to the risk variance for each assumption (pathogen concentrations, water ingestion rate and dose response parameter). In most scenarios and regions, the pathogen concentrations played the largest influence in the risk variability, followed by the ingestion rate. 4. Discussion The results obtained in the present study demonstrated that the surface waters evaluated are heavily contaminated by Giardia cysts and such results can be attributed to the low level of sewage collection and treatment coverage in all regions. The detection of Cryptosporidium in relatively low densities (0.1 to 1.0 oocysts/L), in fewer samples should be considered relevant, due to the high virulence, infectivity and environmental resistance of this parasite. Many Brazilian authors have also reported more elevated percentages of positive samples for Giardia as well as higher densities of this parasite than for Cryptosporidium in surface waters (Cantusio
Table 2 Protozoan parasites monitoring results. Protozoan
Number of samples tested
Giardia VPR MRC MRSP MRSC All regions
cysts/L 48 82 56 20 206
Cryptosporidium VPR MRC MRSP MRSC All regions
oocysts/L 48 82 56 20 206
Number and percentage of samples by protozoan concentration range
Maximum concentration
b0.1
0.1–1,0
1.1–10.0
>10.0
19 (39.6) 36 (43.9) 33 (58.9) 16 (80.0) 104 (50.5)
16 (33.3) 21 (25.6) 8(14.3) 2 (10.0) 47 (22.8)
12 (25.0) 22 (26.8) 11 (19.6) 2 (10.0) 47 (22.8)
1 (2.1) 3 (3.6) 4(7.1) 0 8 (3.9)
22 53 97 1.7 –
0 13 (15.8) 3(5.3) 1 (5.0) 17 (8.2)
0 1 (1.2) 1(1.9) 0 2(1.0)
0 0 0 0 0
b0.1 6 2.5 0.3 –
48 (100,0) 68 (82.9) 52 (92.8) 19 (95.0) 187(90.8)
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
USEPA (1998), in the four regions, except in VPR region, where the risk of infection by Cryptosporidium was not computed due to the absence of positive samples. In general, the mean and median of the risks computed in first scenario (data sets adjusted via maximum likelihood with left-censoring) were similar or slightly smaller than DL and half DL scenarios for all regions. On the other hand, the upper limits of 95% probability intervals for the first scenario tended to be slightly higher than DL and half DL scenarios. This may be explained by the fact that despite mean estimates in the three scenarios were very similar, standard deviation estimates for the adjusted scenario were in general higher than for the DL and half DL scenarios (Table 4). These results are consistent with the conclusions of several authors who discussed the fill-in (replacement) of censored data by constants (esp. DL, half DL or zero). El-Shaarawi and Esterby (1992) showed by analytical methods that, for data with normal or lognormal distribution, conventional estimators of mean and standard deviation with fill-in of missing data by constants are biased, that is, the sample estimated mean and standard deviation may not converge to the true values, independently of the sample size. Fill-in with DL, especially, tends to provide overestimated means and underestimated variances, while fill-in with zeros has an opposite tendency. For this reason, the half DL substitution is considered a “good” balance, or a “neutral” assumption for the unknown values (Gleit, 1985; Govaerts et al., 2005). Maximum likelihood and regression-based methods are considered more consistent and tend to achieve better estimates, as shown empirically by Govaerts et al. (2000) and Haas and Scheff (1990). Nonetheless, as pointed out by Gleit (1985) and Clarke (1998), fill-in with constants may be suitable for small number of samples or when high censoring proportion prevails, because in these cases the most sophisticated methods usually perform poorly. The risk of Giardia was higher than the risk of Cryptosporidium in all the areas and scenarios evaluated (Tables 5 and 6). Ryu and Abbaszadegan (2008) performed a four-year study in the surface waters of central Arizona, US, and also found that the infection risk for Cryptosporidium was lower than that estimated for Giardia. The recovery percentage of the method and the viability rate tested by propidium iodine (PI) were considered to adjust the protozoa dose for risk evaluation. The authors argued that the values of risk (4.9 × 10 − 4–6.0 × 10 − 4) could be overestimated in their study
Table 4 Mean and standard deviation of adjusted gamma distribution concentrations per regions considering the three scenarios. Regions/scenarios
Giardia (cysts/L) All regions VPR MRC MRSP MRSC
Adjusteda
b c
Half DL
Mean
SD
Mean
SD
Mean
SD
1.6464 1.3706 2.1203 1.7864
4.1446 2.7502 4.8869 5.5409
c
c
1.7014 1.4114 2.1466 1.8278 0.2494
2.6729 2.0037 3.3468 3.0670 0.2555
1.6768 1.3941 2.1099 1.8121 0.2088
2.8428 2.1680 3.5661 3.2761 0.2649
0.1601 0.2093 0.1575 0.1099
0.1288 0.2046 0.1236 0.0310
0.1142 0.1684 0.1104 0.0626
0.1205 0.2035 0.1150 0.0316
Cryptosporidium (oocysts/L) All regions 0.0709 MRC 0.1299 c MRSP c MRSC a
DLb
0.3549 0.4637 c c
393
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for this scenario.
Neto and Franco, 2004; Razzolini et al., 2010) and in raw and treated domestic effluents (Cantusio Neto et al., 2006). Carmena et al. (2007) analyzing water samples from rivers and reservoirs in northern Spain found similar results. The frequency of positive results for Giardia cysts varied significantly among the different regions, ranging from 20% to 40.4%. Despite MRSP region having the highest population density, positive sample percentages for Giardia were lower than for MRC and VPR regions. It is important to mention that, in MRSP region, eight of the eleven sampled sites are reservoirs, where protozoa are usually detected in low frequency. On the other hand, the highest concentrations of Giardia were detected in rivers from this region. Although the total number of positive samples for Cryptosporidium oocysts was low, 74% of these positive samples and the highest concentrations of this parasite were detected in MRC region. Further studies are recommended to determine the source of contamination by this protozoan in that region, where land occupation comprises mostly agriculture and cattle breeding. The estimated annual risk of infection by both protozoa for children and adults was superior to the annual risk of 10 −4 issued by
Table 5 Annual probability (mean, median and probability interval limits for 95% of confidence) of Giardia infection by ingestion of drinking water for adults and children. Regions/scenarios
All regions Adjusteda DLb Half DL VPR Adjusted DL Half DL MRC Adjusted DL Half DL MRSP Adjusted DL Half DL MRSC c DL Half DL a b c d e
Adults
Children d
e
Mean
Median
LPI 95%
UPI 95%
Mean
Median
LPI 95%
UPI95%
1.94% 1.99% 1.90%
1.92% 1.98% 1.89%
1.42% 1.65% 1.53%
2.59% 2.38% 2.32%
0.55% 0.57% 0.55%
0.54% 0.56% 0.53%
0.33% 0.40% 0.37%
0.87% 0.81% 0.92%
1.66% 1.69% 1.67%
1.65% 1.68% 1.66%
1.30% 1.40% 1.37%
2.10% 2.03% 2.01%
0.46% 0.50% 0.47%
0.44% 0.49% 0.46%
0.30% 0.34% 0.33%
0.77% 0.72% 0.68%
2.47% 2.48% 2.42%
2.46% 2.47% 2.41%
1.87% 2.03% 1.97%
3.16% 2.98% 2.94%
0.73% 0.73% 0.73%
0.68% 0.70% 0.71%
0.43% 0.50% 0.48%
1.25% 1.27% 1.11%
2.17% 2.18% 2.22%
2.13% 2.17% 2.21%
1.47% 1.77% 1.77%
3.15% 2.64% 2.75%
0.55% 0.66% 0.63%
0.52% 0.63% 0.59%
0.31% 0.44% 0.40%
0.92% 1.07% 1.07%
0.29% 0.24%
0.29% 0.24%
0.25% 0.21%
0.33% 0.28%
0.08% 0.07%
0.08% 0.07%
0.06% 0.05%
0.12% 0.10%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for adjusted scenario. Lower Probability Interval. Upper Probability Interval.
394
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
Table 6 Annual probability (mean, median and probability interval limits for 95% of confidence) of Cryptosporidium infection by ingestion of drinking water for adults and children. Regions/scenarios
All Adjusteda DLb Half DL MRC Adjusted DL Half DL MRSP c DL Half DL MRSCc DL Half DL a b c d e
Adults
Children
Mean
Median
LPId 95%
UPIe 95%
Mean
Median
LPI 95%
UPI 95%
0.10% 0.22% 0.15%
0.10% 0.22% 0.15%
0.05% 0.20% 0.13%
0.17% 0.25% 0.18%
0.02% 0.06% 0.05%
0.02% 0.06% 0.05%
0.01% 0.05% 0.04%
0.05% 0.08% 0.08%
0.18% 0.29% 0.23%
0.18% 0.29% 0.23%
0.11% 0.25% 0.20%
0.26% 0.33% 0.28%
0.05% 0.08% 0.06%
0.04% 0.08% 0.06%
0.02% 0.06% 0.05%
0.09% 0.11% 0.09%
0.22% 0.15%
0.22% 0.15%
0.19% 0.13%
0.24% 0.18%
0.06% 0.05%
0.06% 0.04%
0.05% 0.03%
0.09% 0.06%
0.15% 0.09%
0.15% 0.09%
0.14% 0.08%
0.16% 0.10%
0.04% 0.03%
0.04% 0.02%
0.03% 0.02%
0.05% 0.04%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples are not considered for adjusted scenario. Lower Probability Interval. Upper Probability Interval.
because all species of Giardia and Cryptosporidium were considered for risk calculation, being unlikely that the species detected were pathogenic to humans. During an evaluation of source water performed in Viçosa city (Brazil), Bastos et al. (2011) obtained lower annual infection risks for Cryptosporidium, close to the usually adopted target of 10 − 4. Among the possible factors that could explain the different risk estimates, it can be mentioned that the protozoa were assayed by IFA after concentration by calcium flocculation and membrane filtration, that usually furnishes low recovery rates. Besides that, the water consumption rate was 0.87 L/day based on Australian distribution data, lower than the ingestion rate used in the present study. For these authors, water consumption rate followed by oocyst concentrations and oocyst removal/inactivation played an important effect on infection risk estimates. In contrast, in this study, it was observed that concentration of pathogen followed by water consumption rate were the drivers to the risk estimate (Tables 7 and 8). The water ingestion rate showed larger influence in the second and third scenarios (substitution of
censored data by DL and half DL) than in the first scenario (data sets adjusted via maximum likelihood with left-censoring). This may be explained by the fact that, in the second and third scenarios, fitted concentration distributions tended to have small variances, reducing the influence of this assumption on the estimated risks. Factors such as method recovery, viability, infectivity and immunity influence the risk assessment (Fewtrell et al., 2001; Schets et al., 2008). Fewtrell et al. (2001) pointed out the uncertainties in the risk assessment related to the determination of Cryptostoridium concentration at the various points in the chain of water supply. Schets et al. (2008) in an evaluation study of the recreational water quality of lakes and canals of Amsterdam, discussed the possibility of the infection risks calculated for Cryptosporidium to be overestimated since the results of the viability test employed in their study might not always be correlated with outcomes of in vivo and in vitro infectivity assays. The infection risk reported in the present study may also be overestimated since all cysts detected were considered viable and human infectious. Additionally, it was assumed that adults and children
Table 7 Risk sensitivity indexes for Giardia expressed as contribution to the variance of pathogen concentrations, water ingestion and dose response parameter. Regions/scenarios
Adults Concentration
All regions Adjusteda DLb Half DL VPR Adjusted DL Half DL MRC Adjusted DL Half DL MRSP Adjusted DL Half DL MRSC c DL Half DL a b c
Children Water ingestion
Dose response
Concentration
Water ingestion
Dose response
98.5% 94.2% 95.6%
1.3% 4.3% 3.5%
0.2% 1.5% 0.9%
92.3% 73.3% 78.4%
7.4% 25.6% 20.8%
0.3% 1.1% 0.8%
97.0% 91.8% 93.4%
2.4% 6.2% 5.2%
0.6% 2.0% 1.4%
86.4% 68.3% 73.1%
13.2% 30.6% 26.1%
0.4% 1.1% 0.8%
97.7% 93.4% 95.1%
1.8% 5.0% 3.9%
0.5% 1.6% 1.1%
90.7% 74.6% 78.5%
8.9% 24.7% 20.8%
0.4% 0.7% 0.7%
99.0% 94.2% 96.1%
0.9% 4.3% 3.1%
0.2% 1.5% 0.8%
96.6% 78.0% 81.5%
3.2% 21.3% 17.6%
0.2% 0.7% 0.9%
83.1% 89.3%
13.9% 8.4%
3.0% 2.4%
46.6% 60.3%
51.8% 38.1%
1.6% 1.5%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for adjusted scenario.
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
395
Table 8 Risk sensitivity indexes for Cryptosporidium expressed as contribution to the variance of pathogen concentrations, water ingestion and dose response parameter. Regions/scenarios
All regions Adjusteda DLb Half DL MRC Adjusted DL Half DL MRSP c DL Half DL MRSC c DL Half DL a b c
Adults
Children
Concentration
Water ingestion
Dose response
Concentration
Water ingestion
Dose response
99.8% 68.4% 81.3%
0.2% 20.6% 12.1%
0.0% 11.1% 6.7%
99.2% 31.2% 46.2%
0.8% 64.2% 49.7%
0.0% 4.7% 4.1%
99.3% 77.6% 85.1%
0.6% 15.1% 9.7%
0.1% 7.3% 5.2%
98.0% 42.1% 57.2%
2.0% 53.5% 39.0%
0.1% 4.4% 3.8%
66.9% 80.3%
21.5% 13.0%
11.6% 6.7%
29.6% 46.6%
64.7% 50.1%
5.7% 3.3%
17.5% 42.2%
54.0% 36.3%
28.4% 21.5%
3.5% 12.0%
89.6% 80.9%
6.9% 7.1%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for adjusted scenario.
ingest only unboiled tap water. By the other side the (oo)cyst concentrations were not adjusted according to the recovery rate and secondary transmission was not considered minimizing the excess of risk obtained. The consumption of drinking water per person per day is another key factor for the risk estimation. Although water ingestion rates have been adjusted for Brazilian population (via weighted ingestion rate mean in the age groups) there is an urgent need to carry out a regional survey to obtain a more realistic value of water consumption in these regions and thus minimize the model uncertainties. The rate of acute diarrheic disease (ADD) reported by CVE (2010) is about 1% to 2% and 3% to 7%, for the total population and for children, respectively, in the four regions evaluated (Table 1). The annual risks predicted of Giardia infection in VPR, MRC and MRSP regions for adults and children are consistent with such rate but lower (one order of magnitude) for MRSC region. Zmirou-Navier et al. (2006), using the same dose–response model, also found consistent results between Giardia infection risk and epidemiological data obtained from a daily follow-up of digestive morbidity among 544 volunteers consuming tap water. The estimated risk for children obtained in the present work was lower than the ADD rate, probably due to the use of the same dose– response model for both age groups and lower water ingestion rate for children. However, other exposure routes such as the ingestion of contaminated food, attending daycare centers, playing with contaminated soil and creeks should be taken into account for children. This situation may be more common in low income areas, where besides the lack on health infrastructure, inadequate hygiene, sanitation and environmental conditions prevail. 5. Conclusions The Quantitative Microbiological Risk Assessment conducted to evaluate the safety of drinking water in four densely populated regions of Sao Paulo State, Brazil demonstrated that the infection risks of Giardia and Cryptosporidium are superior to the adopted target of 10 − 4 and emphasizes the need to implement the Water Safety Plans as recommended by WHO (2011). As the majority of the Water Treatment Plants supply cities with more than 100,000 habitants, performance targets for the treatment should be established to achieve the required level of public health risk. Sanitary and health measures need to be implemented urgently to reduce the circulation of these protozoa in the environment. Government policies aiming to improve the urban occupation and the collection and treatment of domestic sewage should be implemented in order to reduce the discharge of raw or poorly treated sewage effluents in source waters.
Acknowledgments Financial support to this research was provided by the Water Resource Fund of Sao Paulo State (FEHIDRO) and the Environmental Company of Sao Paulo State (CETESB).
References APHA. Standard methods for the examination of water and wastewater. 20th edition. Washington DC, USA: APHA, AWWA and WEF; 1998. Bastos RKX, Bevilacqua PD, Andrade RC, Viana DB. Quantitative microbial risk assessment of infection by Cryptosporidium spp. in drinking water from surface water source treated by conventional treatment and disinfection with chlorine dioxide. 16th International Symposium on Health-Related Water Microbiology, Rotorua, New Zealand; 2011. http://www.on-cue.co.nz/HRWM%20CD/Thursday/Bastos.pdf. Cantusio Neto R, Franco RMB. Cryptosporidium spp. and Giardia spp. in different water treatment processes at Campinas, Sao Paulo, Brazil. Higiene Aliment 2004;18:52–9. Cantusio Neto R, Santos JU, Franco RMB. Evaluation of activated sludge treatment and the efficiency of the disinfection of Giardia species cysts and Cryptosporidium oocysts by UV at a sludge treatment plant in Campinas, southeast Brazil. Water Sci Technol 2006;54:89–94. Carmena D, Aguinagalde X, Zigorraga C, Fernández-Crespo JC, Ocio JA. Presence of Giardia cysts and Cryptosporidium oocysts in drinking water supplies in northern Spain. J Appl Microbiol 2007;102:619–29. Carvalho-Almeida TT, Pinto PLS, Quadros CMS, Torres MAGV, Kanamura HY, Casimiro AM. Detection of Cryptosporidium sp. in non-diarrheal faeces from children in a day care center in the city of São Paulo, Brazil. Rev Inst Med Trop Sao Paulo 2006;48(1):27–32. Clarke JU. Evaluation of censored data methods to allow statistical comparisons among very small samples with below detection limit observations. Environ Sci Technol 1998;32:177–83. Craun MF, Craun GF, Calderon RL, Beach MJ. Waterborne outbreaks reported in the United States. J Water Health 2006;4(Suppl. 2):19–30. CVE. Informação sobre Doenças Transmitidas por Água e Alimento. Monitoração de Doenças Diarreicas (MDDA); 2010. http://www.cve.saude.sp.gov.br/htm/hidrica/ dta_mdda.htm. Decisioneering. Crystal ball 2000.2 User manual. Denver, Colorado; 2001. Delignette-Muller ML, Pouillot R, Denis JB, Dutang C. Fitdistrplus: help to fit of a parametric distribution to non-censored or censored data. R package version 0.1–3; 2010. http://cran.r-project.org/web/packages/fitdistrplus/index.html. El-Shaarawi HA, Esterby SR. Replacement of censored observations by a constant: an evaluation. Water Res 1992;26(6):835–44. Fewtrell L, Macgill SM, Kay D, Casemore D. Uncertainties in risk assessment for the determination of drinking water pollutant concentrations: Cryptosporidium case study. Water Res 2001;35(2):441–7. Franco RMB, Cordeiro NSS. Giardiasis and cryptosporidiosis in day-care centers in the municipality Campinas SP. Rev Soc Bras Med Trop 1996;29(6):585–91. Gleit A. Estimation for small normal data sets with detection limits. Environ Sci Technol 1985;19:1201–6. Gonçalves EMN, Silva AJ, Eduardo MBP, Uemura IH, Moura INS, Castilho VLP, Corbett CEP. Multilocus genotyping of Cryptosporidium hominis associated with diarrhea outbreak in a day care unit in São Paulo. Clinics 2006;61(2):119–26. Govaerts BC, Le Bailly C, Eeckaut PV. Calculating summary statistics and fitting a lognormal distribution to left censored environmental data. Consulting Report 0002. Institut de Statistique, UCL; 2000.
396
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
Govaerts B, Beck B, Lecoutre E, Le Bailly C, Eeckaut PV. From monitoring data to regional distributions: a practical methodology applied to water risk assessment. Environmetrics 2005;16:109–27. Haas CH, Scheff PA. Estimation of averages in truncated samples. Environ Sci Technol 1990;24:912–9. Haas CH, Crockett CS, Rose JB, Gerba CP, Fazil AM. Assessing the risk posed by oocysts in drinking water. J Am Wat Works Assoc 1996;88(9):131–6. Kahn H, Stralka K. Estimated daily average per capita water ingestion by child and adult age categories based on USDA's 1994–1996 and 1998 continuing survey of food intakes by individuals. J Exp Sci Environ Epidemiol 2009;19:396–404. Karanis P, Kourenti C, Smith H. Waterborne transmission of protozoan parasites: a worldwide review of outbreaks and lessons learnt. J Water Health 2007;5(1):1-38. Mascarini LM, Donalisio MR. Giadiasis and cyptosporidosis in children institutionalized at daycare centers in the state of São Paulo. Rev Soc Bras Med Trop 2006;39(6): 577–9. Meeker WQ, Escobar LA. Statistical methods for reliability data. New York: Wiley Series in Probability and Statistics; 1998. Nieminski EC. Removal of Cryptosporidium and Giardia trough conventional water treatment and direct filtration. EPA/600/SR-97/025; July 1997. Prado MS, Strina A, Barreto ML, Oliveira-Assis AM, Paz LM, Cairncross S. Risks factors for infection with Giardia duodenalis in pre-school children in the city of Salvador, Brazil. Epidemiol Infect 2003;131:899–906. R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria; 2011. http://www.R-project.org. Razzolini MT, Santos TFS, Bastos VK. Detection of Giardia and Cryptosporidium cysts/oocysts in watersheds and drinking water sources in Brazil urban areas. J Water Health 2010;8(2):399–404.
Rose JB, Haas CN, Regli S. Risk assessment and control of waterborne giardiasis. Am J Public Health 1991;1:709–13. Ryu H, Abbaszadegan M. Long-term study of Cryptosporidium and Giardia occurrence and quantitative microbial risk assessment in surface waters of Arizona in the USA. J Water Health 2008;06(2):263–73. Schets FM, Van Wijnen JH, Schijven JF, Schoon H, de Roda Husman AM. Monitoring of waterborne pathogens in surface waters in Amsterdam, The Netherlands, and the potential health associated with exposure to Cryptosporidium and Giardia in these waters. Appl Environ Microbiol 2008;74(7):2069–78. Teixeira JC, Heller L, Barreto ML. Giardia duodenalis infectious: risk factors for children living in sub-standard settlements in Brazil. Cad Saude Publica 2007;23(6): 1489–93. USEPA. National primary drinking water regulations: interim enhanced surface water treatment rule; final rule. 40 CFR parts 9, 141 and 142. December 16; 1998. USEPA. Method 1623: Cryptosporidium and Giardia in water by filtration/IMS/FA. Washington: Office of Water; 1999. Ware MW, Wymer L, Lindquist A, Schaefer FW. Evaluation of an alternative IMS dissociation procedure for use with Method 1622: detection of Cryptosporidium in water. J Microbiol Methods 2003;55:575–83. WHO. Guidelines for drinking-water quality. Fourth ed. Geneva: WHO; 2011. Zmirou-Navier D, Gofti-Laroche L, Hartemann Ph. Waterborne microbial risk assessment: a population-based dose–response function for Giardia spp. BMC Public Health 2006;6:122. http://www.biomedcentral.com/1471-2458/6/122.
Contents lists available at SciVerse ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Assessing the infection risk of Giardia and Cryptosporidium in public drinking water delivered by surface water systems in Sao Paulo State, Brazil Maria Ines Z. Sato a,⁎, Ana Tereza Galvani a, Jose Antonio Padula a, Adelaide Cassia Nardocci b, Marcelo de Souza Lauretto c, Maria Tereza Pepe Razzolini b, Elayse Maria Hachich a a b c
CETESB — Companhia Ambiental do Estado de Sao Paulo, Av. Prof. Frederico Hermann Jr., 345, São Paulo, SP 05459-900, Brazil Faculdade de Saude Publica, Universidade de Sao Paulo, Av. Dr Arnaldo 715 1o andar, Sao Paulo, SP 01246-904, Brazil EACH — Escola de Artes, Ciencias e Humanidade, Universidade de Sao Paulo, R. Arlindo Bettio, 1000, São Paulo, SP 03828-000, Brazil
H I G H L I G H T S ► ► ► ► ►
Giardia was detected in high densities in source waters of urbanized watersheds. Cryptosporidium was seldom found and the oocysts densities were usually low. Giardia and Cryptosporidium infection risk was higher than 1/10,000. Giardia risk infection was greater for adults than that observed for children. The implementation of Water Safe Plan is a tool needed for ensuring improved water.
a r t i c l e
i n f o
Article history: Received 17 July 2012 Received in revised form 28 September 2012 Accepted 30 September 2012 Available online 22 November 2012 Keywords: Risk assessment Giardia Cryptosporidium Drinking water Source water
a b s t r a c t A survey of Giardia and Cryptosporidium was conducted in surface water used as drinking water sources by public water systems in four densely urbanized regions of Sao Paulo State, Brazil. A Quantitative Microbial Risk Assessment, based on protozoa concentrations, was performed to estimate the probability of protozoa infection associated with drinking water ingestion. A total of 206 source water samples were analyzed over a 24 month period using the USEPA Method 1623. The risk of infection was estimated using an exponential dose response model, children and adults exposure and a gamma distribution for (oo)cyst concentrations with three scenarios for treating censored data. Giardia was detected in 102 of the samples, and 19 of them were also positive for Cryptosporidium, with maximum concentrations of 97.0 cysts/L and 6.0 oocysts/L, respectively. Risk distributions were similar for the three scenarios. In the four regions, the estimated risk of Giardia infection per year, for adults and children, ranged from 0.29% to 2.47% and from 0.08% to 0.70%, respectively. Cryptosporidium risk infection varied from 0.15% to 0.29% for adults and from 0.04% to 0.08% for children. In both cases, the calculated risk surpassed the risk of infection of 10 −4 (1:10,000) defined as tolerable by USEPA for a yearly exposure. The probability of Giardia infection was very close to the rates of acute diarrheic disease for adults (1% to 3%) but lower for children (2% to 7%). The daily consumption of drinking water was an important contributing factor for these differences. The Microbiological Risk Assessment carried out in this study provides an indication of infection risks by Giardia and Cryptosporidium in the population served by these source waters. Strategies for source water protection and performance targets for the water treatment should be established to achieve the required level of public health risk. © 2012 Elsevier B.V. All rights reserved.
1. Introduction Microbiologically unsafe drinking water supplies are still a public health problem all over the world but may be critical for developing ⁎ Corresponding author at: CETESB — Companhia Ambiental do Estado de Sao Paulo, Departamento de Analises Ambientais, Av. Prof. Frederico Hermann Jr., 345, Alto de Pinheiros, Sao Paulo, SP 05459-900, Brazil. Tel.: +55 11 3133 3541; fax: +55 11 3133 3982. E-mail addresses: [email protected], [email protected] (M.I.Z. Sato), [email protected] (A.T. Galvani), [email protected] (J.A. Padula), [email protected] (A.C. Nardocci), [email protected] (M.S. Lauretto), [email protected] (M.T.P. Razzolini), [email protected] (E.M. Hachich). 0048-9697/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2012.09.077
countries, especially in big cities, where the lack of adequate health services, sanitation and safe water combined with high population density and poor quality housing prevail in the peri-urban areas. Many metropolitan regions of Sao Paulo State in Brazil, the richest State of the country, share these features. According to health surveillance data, acute diarrheal diseases incidence is 1.2% in the population and reach 5% in children under 5 years old (CVE, 2010). Drinking water supplies for Sao Paulo State population come predominantly from surface water sources and some of these water bodies are heavily contaminated by discharge of raw or partially treated sewage from nearby towns.
390
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
Among the waterborne pathogens, protozoa pose major challenges to design and maintenance of safe water supplies (Zmirou-Navier et al., 2006). The protozoan parasites Giardia lamblia and Cryptosporidium hominis are recognized as important waterborne disease pathogens and are associated with severe gastrointestinal illness. Craun et al. (2006) reported that these parasites have been the most frequently identified etiologic agents in the last 12 years in the United States. A recent worldwide review of waterborne outbreaks caused by protozoan parasites (Karanis et al., 2007) provides detailed information regarding date and location, estimated number of cases and suspected causes of the outbreaks. According to this survey, out of 325 outbreaks, 32% (104) were associated with drinking water systems contaminated with G. lamblia, and 23.7% (77) had Cryptosporidium parvum or Cryptosporidium sp. as the etiologic agent. Deficiencies in the water treatment processes were the most cited reasons. Such deficiencies included gaps in the protective barriers and poorly operated treatment and disinfection systems. In Brazil, cryptosporidiosis and giardiasis represent an important cause of morbidity in children from 0 to 5 years (Carvalho-Almeida et al., 2006; Franco and Cordeiro, 1996; Gonçalves et al., 2006). Teixeira et al. (2007) reported a prevalence of Giardia in 18% of children in a sub-standard settlement in Juiz de Fora city (Brazil). Prado et al. (2003) in a study about risk factors for Giardia infection in pre-school children (aged 2–45 months) in the city of Salvador (Brazil) found that 13.7% (95/694) of them were infected with Giardia duodenalis. According to Mascarini and Donalisio (2006) there is high prevalence of giardiasis among children who are
attended in daycare centers in Brazil, mainly in those ones located in poor areas. The objectives of this study were to survey the prevalence of the protozoa Giardia and Cryptosporidium in surface water used as drinking water sources by public water systems in four densely populated regions of Sao Paulo State and to assess the risk of protozoan infection for the population supplied with drinking water from these sources. 2. Material and methods 2.1. Sampling The water sampling sites were situated in nine different watersheds at eastern region of the State covering the Metropolitan Region of Sao Paulo (MRSP), Metropolitan Region of Campinas (MRC), Metropolitan Region of South Coast (MRSC) and Vale do Paraiba region (VPR), which together comprise 58% of the Sao Paulo state and 13% of the Brazilian population (Fig. 1). The main characteristics of each region selected are summarized in Table 1. Ten liter volumes of water were collected for Giardia and Cryptosporidium analysis during 24 months (bimonthly) in 28 collection sites located at the source water intake area of water treatment plants (WTP) being five sites at VPR, nine sites at MRC, eleven sites at MRSP and three sites at MRSC. The water treatment processes in all WTPs consisted of coagulation, flocculation, sedimentation, sand filtration, chlorination and fluoridation. The water samples were
Fig. 1. Location of the four regions evaluated in Sao Paulo State, southeast of Brazil.
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396 Table 1 Main characteristics of the study area. VPR Populationa Proportion of state population (%)a Population density (inhab./km2)a Geometric rate of annual population growth (%) a Urbanization rate (%)a Yearly infant mortality rate (per 1000 birth)a Gross Domestic Product (GDP) per capita (USD)a Contribution to state GDP (%)a Drinking water coverage (%)a Sewage collection (%)a Sewage treatment (%)a Acute diarrheic disease rate (b5 years old) (%)b Acute diarrheic disease rate total population (%)b
MRC
MRSP
MRSC
2,014,219 2,847,077 19,847,879 1,682,435 5 7 48 4 141.95 0.83
780.95 1.82
2498.53 0.97
694.42 1.2
76.65 17.32
97.43 10.32
98.77 12.35
99.79 18.80
14,600
14,000
15,000
12,000
5 93.4 83 48 7
8 97.1 82 46 3
57 97.5 83 44 6
4 96.0 67 9 4
2
1
1
2
a Fundação SEADE (2011) — State Foundation of Data Analysis System (http:// www.seade.gov.br/). b CVE (2010) — Acute diarrheic disease rate was calculated from the number of the cases divided by correspondent aged population.
collected according to APHA (1998), kept on ice for transportation and processed within 24 h. 2.2. Protozoan analyses Protozoan parasites were detected using immunomagnetic separation–immunofluorescence assay (IMS-IFA) following USEPA 1623 Method (USEPA, 1999). Water samples were filtered through a 1 μm polycarbonate membrane (Whatman Inc.) and immunomagnetic separation was performed with Dynal reagents and equipment. The dissociation of the complex beads, cysts and oocysts was made by heat dissociation (Ware et al., 2003). The slides were stained using fluorescein conjugated monoclonal antibodies for Giardia and Cryptosporidium (Waterborne) and DAPI (4′, 6-diamidino-2-phenyl-indole, Sigma). The cysts and oocysts were identified and counted by immunofluorescence reaction and confirmed by DAPI fluorescence and DIC (differential interference contrast microscopy) using a Leica DMLD epifluorescence microscope equipped with bright field, phase contrast DIC and epifluorescence optics. Negative and positive control slides were also prepared. Recovery efficiency of the method was determined by spiking 10 L of purified water samples with EasySeed, according to the instructions of the manufacturer (BTFbio). The theoretical detection limit was calculated based on the assumption that at least one cyst or one oocyst could be detected by the microscopic examination of the slides, the entire packed pellet was examined and a 10 L volume of water sample was concentrated; therefore 1 oo(cyst) in 10 L or 0.1 oo(cyst) per liter. 2.3. Risk assessment procedure For conducting the risk assessment, raw water concentrations of Giardia and Cryptosporidium, log10 removal of (oo)cysts of Giardia and Cryptosporidium through the conventional water treatment, the dose–response model and the ingestion rate of water by children (b5 years old) and adults (>21 years old) were taken into account. 2.3.1. Concentrations of Giardia and Cryptosporidium in raw water In each region, concentrations of both organisms were adjusted via standard distributions. According to a preliminary analysis, lognormal and gamma distributions were considered and fitted by the maximum likelihood method. A graphical analysis and a comparison among maximum likelihood values suggested that the gamma distribution achieved
391
a better fitting, being therefore the distribution of choice for the risk analysis. A characteristic of the datasets was the occurrence of several left-censored observations — samples with concentrations below the theoretical detection limit (DL). Due to this reason, three scenarios were employed for treating censored data, as suggested by Clarke (1998) and Govaerts et al. (2005). In the first scenario, datasets were adjusted via maximum likelihood with left-censoring (Meeker and Escobar, 1998; Govaerts et al., 2005), implemented in R Package “fitdistrplus” (Delignette-Muller et al., 2010). Regions with less than five positive samples were not considered for this scenario. For the second scenario, the DL was assumed for censored data (b0.1(oo)cysts/L) whereas for the third scenario half of DL was used. For scenarios 2 and 3, filled-in data were adjusted via traditional (non-censored) gamma distributions. 2.3.2. Log 10 removal by the treatment process The removal efficiency of protozoan by the conventional treatment processes was assumed to be 3.0 log (99.9%) and 2.0 log (99%) for Giardia and Cryptosporidium, respectively as demonstrated by Nieminski (1997). All (oo)cysts were considered viable and infectious. 2.3.3. Dose response model The exponential dose–response model was used for both pathogens. The dose response parameter value (r-organism-specific infectivity parameter) adopted for Giardia was 0.01982 with a 95% Confidence Interval (CI) of 0.009798–0.03582 (Rose et al., 1991). For Cryptosporidium, r value was 0.00467 (95% CI: 0.00097–0.00915) (Haas et al., 1996). 2.3.4. Ingestion rate The water ingestion rates for the southeast region Brazilian population of different age categories (b5 and >21 years old) were calculated based on a research carried out by Kahn and Stralka (2009). A lognormal distribution for water ingestion rate with mean and standard deviation (SD) of 0.44 L/day (SD 0.92 L/day) for children and of 1.5 L/day (SD 0.80 L/day) for adults was assumed. 2.3.5. Risk calculation The daily risk of infection (Pi) was calculated by the equation: Pi ¼ 1− expð−r C IR TRÞ where: r C IR TR
dose–response model parameter; average concentration of Giardia or Cryptosporidium ((oo) cysts/L); ingestion rate (L/day); Log 10 treatment removal (3 logs for Giardia and 2 logs for Cryptosporidium).
Monte Carlo simulation was performed using R (R Development Core Team, 2011), providing 10,000 simulated values of Pi varying r, C and IR. Annual risk of infection (Pa) was computed by drawing 365 values Pi (1), Pi (2),…, Pi (365) from the 10,000 values of Pi, without replacement, and by applying the equation: 365 ðj Þ Pa ¼ 1−∏j¼1 1−Pi : This procedure was repeated 10,000 times to provide the annual risk simulated distribution Pa and the corresponding values of empirical mean, median and 95% probability interval (2.5% and 97.5% quantiles). 2.3.6. Sensitivity analysis In order to estimate the influence of the assumptions r, C and IR on risk results a sensitivity analysis was performed according to
392
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
the method described in Decisioneering (2001) and references therein. First, the rank correlation coefficients between every assumption and every estimated daily risk were computed. Then, the contributions to the variance were calculated by squaring the rank correlation coefficients for the assumptions and normalized to 100%. The contribution to the variance is an approximated method to estimate the percentage of the variance in the risk due to each assumption. 3. Results 3.1. Protozoa detection A total number of 206 samples collected throughout the 24 months, from 28 locations, were tested for both protozoa. Giardia was detected in 49.5% of the samples, in concentrations ranging from 0.1cyst/L to 97 cysts/L. Cryptosporidium was detected in 9.2% of the samples, in concentrations between 0.1 oocyst/L and 6 oocysts/L. Table 2 shows the number and percentage of samples according to the different protozoa concentration ranges. For Giardia most of the positive results were in the range of 0.1 to 10.0 cysts per liter, while for Cryptosporidium these values were in the range of 0.1–1.0 oocysts per liter. Although VPR and MRC regions have presented the highest percentage of positive samples for Giardia, the maximum concentration of this protozoan was detected in the Metropolitan Region of Sao Paulo (MRSP): densities of 97 cysts/L were found in samples collected in two consecutive months from the same waterbody. The highest frequency of positive samples and the highest concentrations of Cryptosporidium oocysts were observed in the Metropolitan Region of Campinas. This protozoan was not detected in any sample from VPR. Initial Precision and Recovery (IPR) performed for the Method 1623 showed recovery percentages of 39.4% for Giardia and of 41.3% for Cryptosporidium. These recovery rates were not used for the risk calculation. The adjusted gamma distribution concentration parameters (shape and rate) of each region and scenarios are shown in Table 3. Table 4 presents the estimated mean and standard deviation for the fitted distribution. Results indicated that, in each region, fitted distributions in the DL scenario have similar means for the three scenarios. In all regions, standard deviations for adjusted scenario were slightly higher than those in DL and half DL scenarios. 3.2. Risk assessment Tables 5 and 6 summarize the annual probability of infection by ingestion of cysts and oocysts for adults and children, computed in each region and scenario. Risk distributions were similar for the three scenarios and showed approximately a symmetric shape, as indicated by their means, medians and probability interval limits. All
Table 3 Parameters of gamma distribution concentrations per regions considering the three scenarios. Regions/scenarios
Giardia (cysts/L) All regions VPR MRC MRSP MRSC
Adjusteda Shape
Rate
0.1574 0.2473 0.1887 0.1058
0.0950 0.1789 0.0901 0.0595
c
c
Cryptosporidium (oocysts/L) All regions 0.0397 MRC 0.0758 c MRSP c MRSC a b c
DLb Shape
0.5485 0.5774 c c
Half DL Rate
Shape
Rate
0.4025 0.4917 0.4046 0.3576 0.9546
0.2369 0.3478 0.1900 0.1948 3.8182
0.3424 0.4198 0.3507 0.3020 0.6220
0.2045 0.3012 0.1664 0.1674 2.9618
1.5361 1.0386 1.6270 12.5429
9.6177 4.9799 10.3536 114.0318
0.8940 0.6706 0.9233 3.9024
7.8207 4.0138 8.3396 62.4335
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for this scenario.
values obtained were above the annual acceptable risk level of 10 −4 (0.01%) recommended by USEPA (1998). The median probability of Giardia infection for all regions was about 2% for adults and 0.6% for children, and the highest risk was estimated for the Metropolitan Region of Campinas, followed by Sao Paulo and Vale do Paraiba Regions. The probability of infection by Cryptosporidium was lower than that by Giardia. Due to the low frequency of positive samples for this parasite, the estimated risks showed the influence of the scenario choice, DL scenario presenting the more conservative results. The highest estimated risk was obtained in MRC region, followed by MRSP and MRSC regions. Tables 7 and 8 present the risk sensitivity indexes expressed as contribution to the risk variance for each assumption (pathogen concentrations, water ingestion rate and dose response parameter). In most scenarios and regions, the pathogen concentrations played the largest influence in the risk variability, followed by the ingestion rate. 4. Discussion The results obtained in the present study demonstrated that the surface waters evaluated are heavily contaminated by Giardia cysts and such results can be attributed to the low level of sewage collection and treatment coverage in all regions. The detection of Cryptosporidium in relatively low densities (0.1 to 1.0 oocysts/L), in fewer samples should be considered relevant, due to the high virulence, infectivity and environmental resistance of this parasite. Many Brazilian authors have also reported more elevated percentages of positive samples for Giardia as well as higher densities of this parasite than for Cryptosporidium in surface waters (Cantusio
Table 2 Protozoan parasites monitoring results. Protozoan
Number of samples tested
Giardia VPR MRC MRSP MRSC All regions
cysts/L 48 82 56 20 206
Cryptosporidium VPR MRC MRSP MRSC All regions
oocysts/L 48 82 56 20 206
Number and percentage of samples by protozoan concentration range
Maximum concentration
b0.1
0.1–1,0
1.1–10.0
>10.0
19 (39.6) 36 (43.9) 33 (58.9) 16 (80.0) 104 (50.5)
16 (33.3) 21 (25.6) 8(14.3) 2 (10.0) 47 (22.8)
12 (25.0) 22 (26.8) 11 (19.6) 2 (10.0) 47 (22.8)
1 (2.1) 3 (3.6) 4(7.1) 0 8 (3.9)
22 53 97 1.7 –
0 13 (15.8) 3(5.3) 1 (5.0) 17 (8.2)
0 1 (1.2) 1(1.9) 0 2(1.0)
0 0 0 0 0
b0.1 6 2.5 0.3 –
48 (100,0) 68 (82.9) 52 (92.8) 19 (95.0) 187(90.8)
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
USEPA (1998), in the four regions, except in VPR region, where the risk of infection by Cryptosporidium was not computed due to the absence of positive samples. In general, the mean and median of the risks computed in first scenario (data sets adjusted via maximum likelihood with left-censoring) were similar or slightly smaller than DL and half DL scenarios for all regions. On the other hand, the upper limits of 95% probability intervals for the first scenario tended to be slightly higher than DL and half DL scenarios. This may be explained by the fact that despite mean estimates in the three scenarios were very similar, standard deviation estimates for the adjusted scenario were in general higher than for the DL and half DL scenarios (Table 4). These results are consistent with the conclusions of several authors who discussed the fill-in (replacement) of censored data by constants (esp. DL, half DL or zero). El-Shaarawi and Esterby (1992) showed by analytical methods that, for data with normal or lognormal distribution, conventional estimators of mean and standard deviation with fill-in of missing data by constants are biased, that is, the sample estimated mean and standard deviation may not converge to the true values, independently of the sample size. Fill-in with DL, especially, tends to provide overestimated means and underestimated variances, while fill-in with zeros has an opposite tendency. For this reason, the half DL substitution is considered a “good” balance, or a “neutral” assumption for the unknown values (Gleit, 1985; Govaerts et al., 2005). Maximum likelihood and regression-based methods are considered more consistent and tend to achieve better estimates, as shown empirically by Govaerts et al. (2000) and Haas and Scheff (1990). Nonetheless, as pointed out by Gleit (1985) and Clarke (1998), fill-in with constants may be suitable for small number of samples or when high censoring proportion prevails, because in these cases the most sophisticated methods usually perform poorly. The risk of Giardia was higher than the risk of Cryptosporidium in all the areas and scenarios evaluated (Tables 5 and 6). Ryu and Abbaszadegan (2008) performed a four-year study in the surface waters of central Arizona, US, and also found that the infection risk for Cryptosporidium was lower than that estimated for Giardia. The recovery percentage of the method and the viability rate tested by propidium iodine (PI) were considered to adjust the protozoa dose for risk evaluation. The authors argued that the values of risk (4.9 × 10 − 4–6.0 × 10 − 4) could be overestimated in their study
Table 4 Mean and standard deviation of adjusted gamma distribution concentrations per regions considering the three scenarios. Regions/scenarios
Giardia (cysts/L) All regions VPR MRC MRSP MRSC
Adjusteda
b c
Half DL
Mean
SD
Mean
SD
Mean
SD
1.6464 1.3706 2.1203 1.7864
4.1446 2.7502 4.8869 5.5409
c
c
1.7014 1.4114 2.1466 1.8278 0.2494
2.6729 2.0037 3.3468 3.0670 0.2555
1.6768 1.3941 2.1099 1.8121 0.2088
2.8428 2.1680 3.5661 3.2761 0.2649
0.1601 0.2093 0.1575 0.1099
0.1288 0.2046 0.1236 0.0310
0.1142 0.1684 0.1104 0.0626
0.1205 0.2035 0.1150 0.0316
Cryptosporidium (oocysts/L) All regions 0.0709 MRC 0.1299 c MRSP c MRSC a
DLb
0.3549 0.4637 c c
393
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for this scenario.
Neto and Franco, 2004; Razzolini et al., 2010) and in raw and treated domestic effluents (Cantusio Neto et al., 2006). Carmena et al. (2007) analyzing water samples from rivers and reservoirs in northern Spain found similar results. The frequency of positive results for Giardia cysts varied significantly among the different regions, ranging from 20% to 40.4%. Despite MRSP region having the highest population density, positive sample percentages for Giardia were lower than for MRC and VPR regions. It is important to mention that, in MRSP region, eight of the eleven sampled sites are reservoirs, where protozoa are usually detected in low frequency. On the other hand, the highest concentrations of Giardia were detected in rivers from this region. Although the total number of positive samples for Cryptosporidium oocysts was low, 74% of these positive samples and the highest concentrations of this parasite were detected in MRC region. Further studies are recommended to determine the source of contamination by this protozoan in that region, where land occupation comprises mostly agriculture and cattle breeding. The estimated annual risk of infection by both protozoa for children and adults was superior to the annual risk of 10 −4 issued by
Table 5 Annual probability (mean, median and probability interval limits for 95% of confidence) of Giardia infection by ingestion of drinking water for adults and children. Regions/scenarios
All regions Adjusteda DLb Half DL VPR Adjusted DL Half DL MRC Adjusted DL Half DL MRSP Adjusted DL Half DL MRSC c DL Half DL a b c d e
Adults
Children d
e
Mean
Median
LPI 95%
UPI 95%
Mean
Median
LPI 95%
UPI95%
1.94% 1.99% 1.90%
1.92% 1.98% 1.89%
1.42% 1.65% 1.53%
2.59% 2.38% 2.32%
0.55% 0.57% 0.55%
0.54% 0.56% 0.53%
0.33% 0.40% 0.37%
0.87% 0.81% 0.92%
1.66% 1.69% 1.67%
1.65% 1.68% 1.66%
1.30% 1.40% 1.37%
2.10% 2.03% 2.01%
0.46% 0.50% 0.47%
0.44% 0.49% 0.46%
0.30% 0.34% 0.33%
0.77% 0.72% 0.68%
2.47% 2.48% 2.42%
2.46% 2.47% 2.41%
1.87% 2.03% 1.97%
3.16% 2.98% 2.94%
0.73% 0.73% 0.73%
0.68% 0.70% 0.71%
0.43% 0.50% 0.48%
1.25% 1.27% 1.11%
2.17% 2.18% 2.22%
2.13% 2.17% 2.21%
1.47% 1.77% 1.77%
3.15% 2.64% 2.75%
0.55% 0.66% 0.63%
0.52% 0.63% 0.59%
0.31% 0.44% 0.40%
0.92% 1.07% 1.07%
0.29% 0.24%
0.29% 0.24%
0.25% 0.21%
0.33% 0.28%
0.08% 0.07%
0.08% 0.07%
0.06% 0.05%
0.12% 0.10%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for adjusted scenario. Lower Probability Interval. Upper Probability Interval.
394
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
Table 6 Annual probability (mean, median and probability interval limits for 95% of confidence) of Cryptosporidium infection by ingestion of drinking water for adults and children. Regions/scenarios
All Adjusteda DLb Half DL MRC Adjusted DL Half DL MRSP c DL Half DL MRSCc DL Half DL a b c d e
Adults
Children
Mean
Median
LPId 95%
UPIe 95%
Mean
Median
LPI 95%
UPI 95%
0.10% 0.22% 0.15%
0.10% 0.22% 0.15%
0.05% 0.20% 0.13%
0.17% 0.25% 0.18%
0.02% 0.06% 0.05%
0.02% 0.06% 0.05%
0.01% 0.05% 0.04%
0.05% 0.08% 0.08%
0.18% 0.29% 0.23%
0.18% 0.29% 0.23%
0.11% 0.25% 0.20%
0.26% 0.33% 0.28%
0.05% 0.08% 0.06%
0.04% 0.08% 0.06%
0.02% 0.06% 0.05%
0.09% 0.11% 0.09%
0.22% 0.15%
0.22% 0.15%
0.19% 0.13%
0.24% 0.18%
0.06% 0.05%
0.06% 0.04%
0.05% 0.03%
0.09% 0.06%
0.15% 0.09%
0.15% 0.09%
0.14% 0.08%
0.16% 0.10%
0.04% 0.03%
0.04% 0.02%
0.03% 0.02%
0.05% 0.04%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples are not considered for adjusted scenario. Lower Probability Interval. Upper Probability Interval.
because all species of Giardia and Cryptosporidium were considered for risk calculation, being unlikely that the species detected were pathogenic to humans. During an evaluation of source water performed in Viçosa city (Brazil), Bastos et al. (2011) obtained lower annual infection risks for Cryptosporidium, close to the usually adopted target of 10 − 4. Among the possible factors that could explain the different risk estimates, it can be mentioned that the protozoa were assayed by IFA after concentration by calcium flocculation and membrane filtration, that usually furnishes low recovery rates. Besides that, the water consumption rate was 0.87 L/day based on Australian distribution data, lower than the ingestion rate used in the present study. For these authors, water consumption rate followed by oocyst concentrations and oocyst removal/inactivation played an important effect on infection risk estimates. In contrast, in this study, it was observed that concentration of pathogen followed by water consumption rate were the drivers to the risk estimate (Tables 7 and 8). The water ingestion rate showed larger influence in the second and third scenarios (substitution of
censored data by DL and half DL) than in the first scenario (data sets adjusted via maximum likelihood with left-censoring). This may be explained by the fact that, in the second and third scenarios, fitted concentration distributions tended to have small variances, reducing the influence of this assumption on the estimated risks. Factors such as method recovery, viability, infectivity and immunity influence the risk assessment (Fewtrell et al., 2001; Schets et al., 2008). Fewtrell et al. (2001) pointed out the uncertainties in the risk assessment related to the determination of Cryptostoridium concentration at the various points in the chain of water supply. Schets et al. (2008) in an evaluation study of the recreational water quality of lakes and canals of Amsterdam, discussed the possibility of the infection risks calculated for Cryptosporidium to be overestimated since the results of the viability test employed in their study might not always be correlated with outcomes of in vivo and in vitro infectivity assays. The infection risk reported in the present study may also be overestimated since all cysts detected were considered viable and human infectious. Additionally, it was assumed that adults and children
Table 7 Risk sensitivity indexes for Giardia expressed as contribution to the variance of pathogen concentrations, water ingestion and dose response parameter. Regions/scenarios
Adults Concentration
All regions Adjusteda DLb Half DL VPR Adjusted DL Half DL MRC Adjusted DL Half DL MRSP Adjusted DL Half DL MRSC c DL Half DL a b c
Children Water ingestion
Dose response
Concentration
Water ingestion
Dose response
98.5% 94.2% 95.6%
1.3% 4.3% 3.5%
0.2% 1.5% 0.9%
92.3% 73.3% 78.4%
7.4% 25.6% 20.8%
0.3% 1.1% 0.8%
97.0% 91.8% 93.4%
2.4% 6.2% 5.2%
0.6% 2.0% 1.4%
86.4% 68.3% 73.1%
13.2% 30.6% 26.1%
0.4% 1.1% 0.8%
97.7% 93.4% 95.1%
1.8% 5.0% 3.9%
0.5% 1.6% 1.1%
90.7% 74.6% 78.5%
8.9% 24.7% 20.8%
0.4% 0.7% 0.7%
99.0% 94.2% 96.1%
0.9% 4.3% 3.1%
0.2% 1.5% 0.8%
96.6% 78.0% 81.5%
3.2% 21.3% 17.6%
0.2% 0.7% 0.9%
83.1% 89.3%
13.9% 8.4%
3.0% 2.4%
46.6% 60.3%
51.8% 38.1%
1.6% 1.5%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for adjusted scenario.
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
395
Table 8 Risk sensitivity indexes for Cryptosporidium expressed as contribution to the variance of pathogen concentrations, water ingestion and dose response parameter. Regions/scenarios
All regions Adjusteda DLb Half DL MRC Adjusted DL Half DL MRSP c DL Half DL MRSC c DL Half DL a b c
Adults
Children
Concentration
Water ingestion
Dose response
Concentration
Water ingestion
Dose response
99.8% 68.4% 81.3%
0.2% 20.6% 12.1%
0.0% 11.1% 6.7%
99.2% 31.2% 46.2%
0.8% 64.2% 49.7%
0.0% 4.7% 4.1%
99.3% 77.6% 85.1%
0.6% 15.1% 9.7%
0.1% 7.3% 5.2%
98.0% 42.1% 57.2%
2.0% 53.5% 39.0%
0.1% 4.4% 3.8%
66.9% 80.3%
21.5% 13.0%
11.6% 6.7%
29.6% 46.6%
64.7% 50.1%
5.7% 3.3%
17.5% 42.2%
54.0% 36.3%
28.4% 21.5%
3.5% 12.0%
89.6% 80.9%
6.9% 7.1%
Adjusted — values adjusted via maximum likelihood with left-censoring. DL — the theoretical detection limit value was assumed for censored data. Regions with less than five positive samples were not considered for adjusted scenario.
ingest only unboiled tap water. By the other side the (oo)cyst concentrations were not adjusted according to the recovery rate and secondary transmission was not considered minimizing the excess of risk obtained. The consumption of drinking water per person per day is another key factor for the risk estimation. Although water ingestion rates have been adjusted for Brazilian population (via weighted ingestion rate mean in the age groups) there is an urgent need to carry out a regional survey to obtain a more realistic value of water consumption in these regions and thus minimize the model uncertainties. The rate of acute diarrheic disease (ADD) reported by CVE (2010) is about 1% to 2% and 3% to 7%, for the total population and for children, respectively, in the four regions evaluated (Table 1). The annual risks predicted of Giardia infection in VPR, MRC and MRSP regions for adults and children are consistent with such rate but lower (one order of magnitude) for MRSC region. Zmirou-Navier et al. (2006), using the same dose–response model, also found consistent results between Giardia infection risk and epidemiological data obtained from a daily follow-up of digestive morbidity among 544 volunteers consuming tap water. The estimated risk for children obtained in the present work was lower than the ADD rate, probably due to the use of the same dose– response model for both age groups and lower water ingestion rate for children. However, other exposure routes such as the ingestion of contaminated food, attending daycare centers, playing with contaminated soil and creeks should be taken into account for children. This situation may be more common in low income areas, where besides the lack on health infrastructure, inadequate hygiene, sanitation and environmental conditions prevail. 5. Conclusions The Quantitative Microbiological Risk Assessment conducted to evaluate the safety of drinking water in four densely populated regions of Sao Paulo State, Brazil demonstrated that the infection risks of Giardia and Cryptosporidium are superior to the adopted target of 10 − 4 and emphasizes the need to implement the Water Safety Plans as recommended by WHO (2011). As the majority of the Water Treatment Plants supply cities with more than 100,000 habitants, performance targets for the treatment should be established to achieve the required level of public health risk. Sanitary and health measures need to be implemented urgently to reduce the circulation of these protozoa in the environment. Government policies aiming to improve the urban occupation and the collection and treatment of domestic sewage should be implemented in order to reduce the discharge of raw or poorly treated sewage effluents in source waters.
Acknowledgments Financial support to this research was provided by the Water Resource Fund of Sao Paulo State (FEHIDRO) and the Environmental Company of Sao Paulo State (CETESB).
References APHA. Standard methods for the examination of water and wastewater. 20th edition. Washington DC, USA: APHA, AWWA and WEF; 1998. Bastos RKX, Bevilacqua PD, Andrade RC, Viana DB. Quantitative microbial risk assessment of infection by Cryptosporidium spp. in drinking water from surface water source treated by conventional treatment and disinfection with chlorine dioxide. 16th International Symposium on Health-Related Water Microbiology, Rotorua, New Zealand; 2011. http://www.on-cue.co.nz/HRWM%20CD/Thursday/Bastos.pdf. Cantusio Neto R, Franco RMB. Cryptosporidium spp. and Giardia spp. in different water treatment processes at Campinas, Sao Paulo, Brazil. Higiene Aliment 2004;18:52–9. Cantusio Neto R, Santos JU, Franco RMB. Evaluation of activated sludge treatment and the efficiency of the disinfection of Giardia species cysts and Cryptosporidium oocysts by UV at a sludge treatment plant in Campinas, southeast Brazil. Water Sci Technol 2006;54:89–94. Carmena D, Aguinagalde X, Zigorraga C, Fernández-Crespo JC, Ocio JA. Presence of Giardia cysts and Cryptosporidium oocysts in drinking water supplies in northern Spain. J Appl Microbiol 2007;102:619–29. Carvalho-Almeida TT, Pinto PLS, Quadros CMS, Torres MAGV, Kanamura HY, Casimiro AM. Detection of Cryptosporidium sp. in non-diarrheal faeces from children in a day care center in the city of São Paulo, Brazil. Rev Inst Med Trop Sao Paulo 2006;48(1):27–32. Clarke JU. Evaluation of censored data methods to allow statistical comparisons among very small samples with below detection limit observations. Environ Sci Technol 1998;32:177–83. Craun MF, Craun GF, Calderon RL, Beach MJ. Waterborne outbreaks reported in the United States. J Water Health 2006;4(Suppl. 2):19–30. CVE. Informação sobre Doenças Transmitidas por Água e Alimento. Monitoração de Doenças Diarreicas (MDDA); 2010. http://www.cve.saude.sp.gov.br/htm/hidrica/ dta_mdda.htm. Decisioneering. Crystal ball 2000.2 User manual. Denver, Colorado; 2001. Delignette-Muller ML, Pouillot R, Denis JB, Dutang C. Fitdistrplus: help to fit of a parametric distribution to non-censored or censored data. R package version 0.1–3; 2010. http://cran.r-project.org/web/packages/fitdistrplus/index.html. El-Shaarawi HA, Esterby SR. Replacement of censored observations by a constant: an evaluation. Water Res 1992;26(6):835–44. Fewtrell L, Macgill SM, Kay D, Casemore D. Uncertainties in risk assessment for the determination of drinking water pollutant concentrations: Cryptosporidium case study. Water Res 2001;35(2):441–7. Franco RMB, Cordeiro NSS. Giardiasis and cryptosporidiosis in day-care centers in the municipality Campinas SP. Rev Soc Bras Med Trop 1996;29(6):585–91. Gleit A. Estimation for small normal data sets with detection limits. Environ Sci Technol 1985;19:1201–6. Gonçalves EMN, Silva AJ, Eduardo MBP, Uemura IH, Moura INS, Castilho VLP, Corbett CEP. Multilocus genotyping of Cryptosporidium hominis associated with diarrhea outbreak in a day care unit in São Paulo. Clinics 2006;61(2):119–26. Govaerts BC, Le Bailly C, Eeckaut PV. Calculating summary statistics and fitting a lognormal distribution to left censored environmental data. Consulting Report 0002. Institut de Statistique, UCL; 2000.
396
M.I.Z. Sato et al. / Science of the Total Environment 442 (2013) 389–396
Govaerts B, Beck B, Lecoutre E, Le Bailly C, Eeckaut PV. From monitoring data to regional distributions: a practical methodology applied to water risk assessment. Environmetrics 2005;16:109–27. Haas CH, Scheff PA. Estimation of averages in truncated samples. Environ Sci Technol 1990;24:912–9. Haas CH, Crockett CS, Rose JB, Gerba CP, Fazil AM. Assessing the risk posed by oocysts in drinking water. J Am Wat Works Assoc 1996;88(9):131–6. Kahn H, Stralka K. Estimated daily average per capita water ingestion by child and adult age categories based on USDA's 1994–1996 and 1998 continuing survey of food intakes by individuals. J Exp Sci Environ Epidemiol 2009;19:396–404. Karanis P, Kourenti C, Smith H. Waterborne transmission of protozoan parasites: a worldwide review of outbreaks and lessons learnt. J Water Health 2007;5(1):1-38. Mascarini LM, Donalisio MR. Giadiasis and cyptosporidosis in children institutionalized at daycare centers in the state of São Paulo. Rev Soc Bras Med Trop 2006;39(6): 577–9. Meeker WQ, Escobar LA. Statistical methods for reliability data. New York: Wiley Series in Probability and Statistics; 1998. Nieminski EC. Removal of Cryptosporidium and Giardia trough conventional water treatment and direct filtration. EPA/600/SR-97/025; July 1997. Prado MS, Strina A, Barreto ML, Oliveira-Assis AM, Paz LM, Cairncross S. Risks factors for infection with Giardia duodenalis in pre-school children in the city of Salvador, Brazil. Epidemiol Infect 2003;131:899–906. R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria; 2011. http://www.R-project.org. Razzolini MT, Santos TFS, Bastos VK. Detection of Giardia and Cryptosporidium cysts/oocysts in watersheds and drinking water sources in Brazil urban areas. J Water Health 2010;8(2):399–404.
Rose JB, Haas CN, Regli S. Risk assessment and control of waterborne giardiasis. Am J Public Health 1991;1:709–13. Ryu H, Abbaszadegan M. Long-term study of Cryptosporidium and Giardia occurrence and quantitative microbial risk assessment in surface waters of Arizona in the USA. J Water Health 2008;06(2):263–73. Schets FM, Van Wijnen JH, Schijven JF, Schoon H, de Roda Husman AM. Monitoring of waterborne pathogens in surface waters in Amsterdam, The Netherlands, and the potential health associated with exposure to Cryptosporidium and Giardia in these waters. Appl Environ Microbiol 2008;74(7):2069–78. Teixeira JC, Heller L, Barreto ML. Giardia duodenalis infectious: risk factors for children living in sub-standard settlements in Brazil. Cad Saude Publica 2007;23(6): 1489–93. USEPA. National primary drinking water regulations: interim enhanced surface water treatment rule; final rule. 40 CFR parts 9, 141 and 142. December 16; 1998. USEPA. Method 1623: Cryptosporidium and Giardia in water by filtration/IMS/FA. Washington: Office of Water; 1999. Ware MW, Wymer L, Lindquist A, Schaefer FW. Evaluation of an alternative IMS dissociation procedure for use with Method 1622: detection of Cryptosporidium in water. J Microbiol Methods 2003;55:575–83. WHO. Guidelines for drinking-water quality. Fourth ed. Geneva: WHO; 2011. Zmirou-Navier D, Gofti-Laroche L, Hartemann Ph. Waterborne microbial risk assessment: a population-based dose–response function for Giardia spp. BMC Public Health 2006;6:122. http://www.biomedcentral.com/1471-2458/6/122.