The intervention study was conducted in 4 provinces in the Red River Delta of northern Vietnam (Thai Binh, Nam Dinh, Ninh Binh, and Thanh Hoa Provinces) (Figure 1). These provinces were selected because of their known high prevalence of FZT infections among humans, animals, and cultured fish (8,9). The study design was a parallel group design of nurseries: 1 intervention group and 1 nonintervention group. Nurseries were randomly selected from a list provided by the Department of Agriculture and Rural Development of each province. The selection criteria were as follows: small farms (<1,500 m² of ponds), farms managed by household members, and nursery production of rohu (Labeo rohita) in an integrated fish–livestock system. Only 1 pond (study unit) per farm was selected for each intervention and nonintervention nursery. The 2 groups of nurseries were matched as closely as possible by type of production system, location, and size. The fry supplied from hatcheries to the nurseries were confirmed free of FZT metacercariae. Rohu, the fish species produced in all selected nurseries is a well-documented host for FZTs (15).

In 2009, a total of 18 intervention nurseries and 18 nonintervention nurseries from the 4 provinces were included in the study. The intervention trial period was May–September 2009 and involved 1 production cycle of juvenile fish. Studies have shown this to be a period of high FZT transmission (15). In 2010, different groups of 6 intervention and 6 nonintervention nursery ponds from Nam Dinh and Ninh Binh Provinces were added to the trial (for a total of 24 in each group), which was conducted during May–September 2010.

The interventions were based on previously identified risk factors for FZT transmission in aquaculture (Figure 2) (13,17). In 2009 and 2010, we evaluated the combined effects of the interventions aimed at reducing environmental contamination with FZT eggs and reducing snail populations in ponds. All farmers for the nurseries selected for intervention participated in a training course on FZT biology and epidemiology. The intervention study objectives and design were explained, and the farmers were given specific instructions on the planned interventions for their individual nursery, determined on the basis of evaluations of risk factors associated with their nurseries. All farmers in the intervention group provided detailed information about the management practices they used to suppress snail populations and their habitat when preparing the nursery ponds before stocking and about the continuing management of the pond after stocking of fry.

At the start of the trial, the ponds at the intervention nurseries were emptied of water and dried for 5 days, during which time the top 3–5 cm of mud from the pond floor was removed to reduce the number of snails. Farmers were also instructed to remove all vegetation from the pond to reduce snail habitat. The banks of the ponds were lined with concrete to help control vegetation. Furthermore, 2 days before the ponds were refilled with water, aquatic vegetation within 3 m of each side of the water intake outside the pond was removed. A filter sock with a 5-mm mesh was fitted around the water inlet pipe to help reduce snail immigration into the pond.

To assess the effects of snail-control interventions, snail density was estimated before fish were stocked and during the nursery cycle. Samples were obtained in a scoop net, 0.25 m wide, cast 2 m off shore, and dragged perpendicularly back to the pond bank. Thus, the area sampled was 0.25 m × 2 m. A total of 5 such samples per pond were obtained. The samples were washed, and all live snails were brought to a field laboratory, where they were examined for cercarial infection according to procedures described by Dung et al. (14). Snail sampling was conducted before the intervention cycle and at 3, 6, and 9 weeks after stocking of fry during May–September of 2009 and 2010. Snails were identified by using taxonomic keys (18,19). Only snails that were potential hosts of FZTs were included in the analysis, i.e., species of the families Thiaridae (Melanoides tuberculata, Sermyla tornatella, Thiara scabra, and Tarebia granifera) and Bithynidae (mainly Bithynia fuchsiana and Parafossarulus striatulus).

Before the trials, all residents, workers, and immediate neighbors of the intervention nurseries were examined for trematode infections by medically trained staff. The Kato–Katz fecal egg count method was used (20). Two Kato–Katz smears of 41.7-mg fecal samples were independently examined under a microscope at 400× magnification. FZT eggs were counted and recorded as eggs per gram. All persons in whom trematode eggs were found were treated with praziquantel at 25 mg/kg, administered 3×/d for 1 day (total 75 mg/kg). During the 2009 and 2010 intervention trials, fecal examinations were repeated after 30 and 60 days. Residents and workers of nonintervention nurseries were examined for FZT infection at the end of the trials and offered free treatment with praziquantel if found to be infected with FZTs.

Domestic animals (dogs, cats, and pigs) from the households of the intervention nursery workers and the neighboring households were examined for FZT eggs according to the fecal examination procedure described by Lan Anh et al. (21). Animals in which eggs were found were given 1 dose of praziquantel at 40 mg/kg (22,23). Fecal examinations were repeated every 30 days during the production cycle, and animals with positive results were again treated.

Prevention of FZT egg contamination of the nursery ponds consisted of constructing 10–15 cm elevated cement barrier walls along the pond banks to keep surface water runoff from entering the pond. A 50-cm-high fine mesh net was placed around the ponds to keep dogs and cats out. In all intervention nurseries, deliberate discharge of liquid wastes from pigpens and human latrines into the pond as fertilizer was terminated by blocking drains with cement and bricks. During training sessions, farmers were instructed to deposit pig manure and human excreta fertilizer away from the pond to prevent fecal contamination from surface runoff water.

In addition to providing treatment for FZT- infected humans and animals, we also cautioned farmers and members of intervention households to not eat raw or inadequately cooked fish and to not feed uncooked fish to dogs, cats, or pigs. For intervention nurseries, a project team had regular contact at least weekly with the farmers during pond preparation and during the nursery production cycle. During these visits, they discussed experiences and answered questions about the interventions.

Farmers of nonintervention nurseries were informed that a survey was being conducted on nursery management practices and that examination of their juvenile fish was needed for the survey. These farmers were not asked to alter any of their routine practices during the study periods. At the end of the project, all household members from nonintervention nurseries were informed about FZTs and the interventions to reduce FZT transmission.

Before stocking of fry, 100 fry from the supplying hatchery were analyzed for metacercariae by the compression method (24). At 3, 6, and 9 weeks after stocking of the fry, 100 juvenile fish from each pond were randomly collected in a seine net. The fish were placed on ice and transported within 36 h to the laboratory, where they were maintained at 4°C for a maximum of 5 days until processed. Each whole fish was then ground with a small handheld homogenizer and transferred to a beaker for digestion in a pepsin solution (8 mL concentrated HCl and 6 g pepsin in 1,000 mL water) to release metacercariae (17). After digestion, the metacercariae were isolated by sediment washing and identified and counted by examination with a microscope. For each juvenile fish, the number of metacercariae was determined and expressed as intensity (total metacercariae/fish). Up to 10 randomly selected metacercariae from each infected fish were identified to the species level, and the rest of the metacercariae were sorted into >1 type groups: liver flukes (Opisthorchiidae), intestinal flukes (Heterophyidae), and nonzoonotic or dead metacercariae. Metacercariae were identified by using morphologic criteria (11,17). All metacercaria species recovered were pooled for statistical analyses.

Data were double-entered into an Access database (Microsoft Corporation, Redmond, WA, USA) and analyzed by using statistical analysis system software (STATA 11; StataCorp LP, College Station, TX, USA). In 2009, we excluded 1 intervention and 2 nonintervention nursery ponds from the analysis because of high numbers of juvenile fish. In 2010, we excluded another 2 intervention nurseries because the farmers failed to follow intervention protocols.

The number of juvenile fish examined for metacercariae and the area (m²) searched for snails were used as offsets for the statistical analysis. The analysis used was negative binomial regression (25) in a generalized linear model (log link function) among sampling times; interventions and the interaction between sampling times and interventions were entered as factors adjusting for clustering within ponds. The ancillary parameter (α) of the variance function (var = μ + α μ²) was estimated in a full maximum-likelihood estimation (25). The total number of infected juvenile fish was analyzed by using logistic regression with the same predictor as for metacercaria counts and adjusting for clustering within ponds. Furthermore, the numbers of host snails, infected host snails, and relative egg contamination potential were tested as predictors for infections in fish and snails. The relative egg contamination potential for the animals studied was estimated; i.e., actual fecal egg counts in humans, dogs, cats, and pigs were multiplied by the estimated daily feces production, which was 160, 99, 20, and 1,516 g, respectively (21,26). The egg numbers were summed for all humans and animals within a given household.

A total of 9,266 juvenile fish in 2009 and 5,911 in 2010 from intervention and nonintervention nurseries were examined for FZT metacercariae. The following FZT species were identified: Clonorchis sinensis (Opisthorchiidae), Centrocestus formosanus, Haplorchis pumilio (Heterophyidae), and Haplorchis yokogawi (Heterophyidae). Among the recovered metacercariae, 94.00% belonged to the family Heterophyidae, 0.05% were C. sinensis, and 5.95% could not be identified.

During 2009, prevalence of FZT infection increased over time among juvenile fish from intervention and nonintervention ponds (Figure 3); however, the increase was significantly less for intervention farms (p<0.001). The difference in prevalence between intervention and nonintervention nursery ponds increased over time as shown by the significant interaction term (p<0.01) between intervention and time of sampling (Figure 3). At 3 weeks after stocking, the odds of infection in the intervention group were 0.77 (95% CI 0.19–3.12) of that in the nonintervention group; after 6 weeks, odds were 0.47 (95% CI 0.08–2.72); after 9 weeks, odds were only 0.22 (95% CI 0.04–1.14). The intensity of metacercariae infection increased over time (p<0.001) and was significantly lower among juvenile fish from the intervention nurseries (p<0.05) (Figure 4). The difference in intensity of metacercariae between intervention and nonintervention nurseries increased over time, and the interaction between the 2 types of nursery was significant (p<0.01). Thus, at 3 weeks after stocking, the intensity of infection in the intervention group was 0.75 (95% CI 0.14–3.82; p>0.05) of that in the nonintervention group; and after 6 and 9 weeks, the intensity was 0.11 (95% CI 0.02–0.68; p<0.05) and 0.19 (0.04–0.87; p<0.05), respectively.

In 2010, a similar pattern developed during the production cycle. At 3 weeks after stocking, the odds of infection in the intervention group were 0.24 (95% CI 0.05–1.23) of that in the nonintervention group, and after 6 and 9 weeks these odds were 0.35 (95% CI 0.07–1.79) and 0.62 (95% CI 0.12–3.19), respectively. Prevalence of infection increased considerably between 6 weeks and 9 weeks at intervention and nonintervention nurseries (Figure 3), consequently, the interaction between intervention and time of sampling was not significant.

In 2010, intensity of infection also increased over time (p<0.001) and the interaction term between intervention and time was significant (p<0.01). Thus, at 3 weeks after stocking, intensity of infection in the intervention group was 0.19 (95% CI 0.03–1.01; p>0.05) of that in the nonintervention group, and at 6 weeks it was 0.13 (95% CI 0.03–0.47; p<0.05). However, at 9 weeks, the intensity in juveniles from intervention nurseries increased to 0.47 of that in the nonintervention nurseries (95% CI 0.10–2.29; p<0.05) (Figure 4).

During 2009–2010, a total of 30,590 snails were collected. The potential intermediate host of FZTs, i.e., snails of the families Thiaridae and Bithynidae, accounted for 57% (n = 17,478) and 0.04% (n = 133), respectively, of total collected snails. During these years, density of potential host snails (snails/m²) did not differ between intervention and nonintervention nurseries (Figure 5). The density of host snails was higher in the intervention group than in the nonintervention group when we adjusted for preintervention counts; differences were not significant when this adjustment was not made. However, for 2009, the odds of finding infection in host snails in the intervention group were 0.14 (95% CI 0.07–0.27; p<0.001) of that in the nonintervention group, and the interaction between time and treatment was not significant. For 2010, a significant interaction between time and treatment was found (p<0.05), but odds ratios and 95% CIs between intervention and nonintervention nurseries were <0.01 at all 3 sampling times (p<0.001).

In 2009, providing treatment for humans in the farm households before and during the interventions reduced the prevalence of FZT eggs from 14.8% to 3.2% (Table); prevalence among comparable members of nonintervention households, who were tested at the end of the production cycle, was 14.2%. In 2010, FZT infections in humans were low or absent among members of intervention households and 12.5% among members of nonintervention households (Table).

In 2009, FZT prevalence and density among pigs from intervention and nonintervention nurseries were low (0.8% and <0.5 eggs/g, respectively), and no significant differences were found between intervention nurseries at the start and nonintervention nurseries at the end of the production cycle (Table). Because of the low prevalence among pigs, these data were not included in the 2010 analyses. During 2009, prevalence of FZT eggs among cats decreased from 50.0% to 15.4% and among dogs from 18.8% to 15.6%. In 2010, prevalence among cats decreased from 57.1% to 0 and among dogs, from 11.8% to 6.7% (Table).

Calculations of mean relative contamination index (no. eggs/household/day) for humans, cats, dogs, and pigs were based on total volumes of excreted feces and the actual egg intensities estimated (Table). Total potential egg contamination before the intervention did not differ significantly between intervention and nonintervention households. Total egg contamination potential did not correlate with prevalence and intensity of metacercaria infection in juvenile fish or with prevalence of infection in host snails.