The third nationwide schistosomiasis sampling survey covered all 7 schistosome-endemic provinces (Anhui, Hubei, Hunan, Jiangsu, Jiangxi, Sichuan, and Yunnan). The sampling unit was the administrative village (i.e., basic level of administration, often comprising >1 natural village), with only those villages selected where transmission was ongoing.

A stratified cluster random sampling technique with 3 strata was used for village selection. The 7 schistosome-endemic provinces represented the first stratum. Within each province, the second stratum was categorized by characteristics of environmental ecosystems defined and widely used by Chinese health workers, which includes 8 subtypes: 1) fork beach, 2) islet without embankment, 3) islet with embankment, 4) inner embankment in the lake and marshland region, 5) plateau, 6) mountain, 7) hill in the hilly and mountainous region, and 8) waterway networks in the plain region (20). Within each ecosystem, estimated local prevalence of S. japonicum, which was based on results of recent parasitologic examinations, served as the third stratum and used cut-off prevalences of 1%, 5%, and 10%. In each disease-endemic province, ≈1% of administrative villages of the same prevalence class and ecosystem type were randomly selected. One administrative village was randomly selected if the total number of villages was <100 (Figure 1).

All residents 6–65 years of age in selected villages were invited to participate in the study. If the total number of eligible persons was <1,000, residents from neighboring villages with similar ecoepidemiologic characteristics were recruited until >1,000 persons were included.

All participants were screened for S. japonicum–specific immunoglobulin G by using a standardized ELISA (Shenzhen Kangbaide Biotech Co., Shenzhen, People’s Republic of China). Seropositive persons were tested by stool examination. One stool specimen was obtained from each participant, and 3 Kato-Katz thick smear slides were prepared and examined under a light microscope by experienced laboratory technicians (21). The number of S. japonicum eggs was counted on each slide, and the arithmetic mean value was calculated for each person.

In a subsample of villages (n = 25), study participants were interviewed regarding previous infection with S. japonicum and previous treatment history. Common symptoms, e.g., abdominal pain and diarrhea, were investigated by using a recall period of 2 weeks. Liver and spleen enlargement and the degree of hepatic fibrosis were assessed with a portable ultrasound device. Assessment of illness was performed on supine persons during baseline respiration who had fasted. The fibrotic degree of liver parenchyma was graded between 0 and 3 according to criteria of the World Health Organization–sponsored Cairo Working Group (22,23).

In 11 villages selected at random from different ecosystems and for different infection levels, seropositive study participants provided 1 stool specimen and 3 Kato-Katz thick smear slides were prepared. Miracidium hatching test after concentration of eggs with a nylon tissue bag was used. Persons with positive results in the miracidium hatching test or with eggs on the slides were identified as having S. japonicum infections and served as the standard for the diagnosis (24–26). Comparisons of results of 2 diagnostic approaches enabled calculation of sensitivity of the Kato-Katz technique.

Samples of every new batch of ELISA kits used in the survey were randomly selected and tested with standard serum samples to determine their sensitivity and cross-reactivity with antibodies against hepatitis B virus and Paragonimus spp. Rates of false-negative results of serologic tests (compared with the Kato-Katz method) and Kato-Katz thick smears (compared with the hatching test) were calculated.

All data were checked for internal consistency before double entry into standardized databases at provincial institutes. Senior scientists from IPD validated data and established a masterfile from which coverage, compliance, prevalence, and intensity of S. japonicum infection and total estimates of persons infected were calculated. Statistical analyses were performed by using SAS version 8.0 (SAS Institute Inc., Cary, NC, USA). The following formulas were used to calculate the corrected human infection rate in the village: infection rate p = (x/X)/(1 – Q) × (f/F)/(1 – R) and variance S²_p = p(1 – p)/(X – 1) (1 – X/n), where p is the corrected human infection rate in a sampled village, S²_p is the variance of p, x is the number of the seropositive persons, X is the number of eligible persons screened by the serologic test, f is the number of stool-positive persons by the Kato-Katz technique, F is the number of persons tested by the Kato-Katz technique, Q is the false-negative rate of the serologic test, R is the false-negative rate of Kato-Katz thick smears, and n is the total population in a sampled village.

The χ² test was used to compare S. japonicum infection rates between provinces, ecosystems, occupational groups, sex and age. Ordinal logistic regression analysis was used to investigate whether there was an association between serologic results and the degree of liver fibrosis, stratified by ecosystem.

A rigorous quality control system was implemented throughout the study. Blood samples and Kato-Katz thick smear slides from each survey were kept, and 5% of the samples were randomly selected for re-evaluation in the respective provincial institutes. Surveys were repeated in villages where quality control showed a sensitivity <90%. All original data were entered twice, and 10% of original data were compared with submitted databases at IPD. Data from sites where level of agreement was <95% were re-entered.

Overall, 17,542 administrative villages with an estimated population of 29,059,194 were classified as endemic for S. japonicum in 2004. Appendix Table 1 summarizes the number of villages and persons stratified by different levels of endemicity. Prevalence of S. japonicum infection was <1% in >50% of villages (n = 9,243) inhabited by >15 million persons. A total of 1,334 villages (7.6%) were classified as settings where infection prevalence was >10%. Estimated population size in these settings was 2,059,211 (7.1%).

The present survey was conducted in 239 villages and included 291,167 persons 6–65 years of age. This corresponds to 1.4% of all schistosome-endemic villages and 1.0% of the population living therein. Compliance to undergo a serodiagnostic test and, among S. japonucum–positive persons, to submit a stool sample was high (86.2%).

Overall, 30,680 (12.2%) of 250,987 participants were positive for S. japonucum by ELISA. Of these persons, 94.2% submitted a stool specimen and 9.1% of these specimens were egg positive. Estimated corrected S. japonicum prevalence, when sensitivity of the Kato-Katz method in S. japonicum–endemic villages was taken into account, was 2.5%. Prevalence varied considerably by province, from 0.3% (Jiangsu) to 4.2% (Hunan). Prevalence rates above average were found in Hubei (3.8%) and Jianxi (3.1%) Provinces, and prevalence rates in Anhui, Yunnan, and Sichuan Provinces were 2.2%, 1.7% and 0.9%, respectively.

Stratification by ecosystem showed that the highest corrected prevalence was found in the lake and marshland region (3.8%). Overall prevalence in the hilly and mountainous region was 1.1%. Human infection rates among different subtypes of disease-endemic areas showed that the highest rate (7.1%) was seen in the subtype mountain regions, followed by subtypes islets with embankment (4.3%), inner embankment (4.3%), fork beach (3.4%), and islet without embankment (2.7%). Lower prevalence rates were observed in subtypes hill (1.01%) and plateau (0.3%) in the hilly and mountainous regions. A lower prevalence rate of 0.06% was found in the plain region characterized by waterway networks.

As shown in Figure 2, prevalence of S. japonicum infection in male study participants (2.6%) was higher than that in female participants (2.2%). There was a tendency for prevalence to increase with age, with the highest prevalence found in men 40–50 years of age. Prevalence of infection also varied with occupation. The prevalence in fishermen and boatmen was 3.3%, which was significantly higher than in other professional groups (p<0.001). However, no significant association was found between infection prevalence and educational level (Table).

On the basis of survey results, total number of persons infected with S. japonicum in China in 2004 was estimated to be 726,112 (95% confidence interval [CI] 714,497–737,728). More than 82% of infected persons lived in lake and marshland regions (596,599, 95% CI 585,910–607,287), and most of the remaining persons resided in hilly and mountainous regions (128,720, 95% CI 124,206–133,233).

The geometric mean of infection intensity was 33 eggs per gram (EPG) of feces among egg-positive persons and 0.4 EPG in the general population. Stratified by province, the highest geometric mean among egg-positive persons was found in Jiangxi (56 EPG) and Hunan (0.70 EPG) when the general population was considered. With regard to ecosystem stratification, the highest infection intensity in egg-positive persons was in the plain region with waterway networks (128 EPG), and the highest infection intensity in the entire population was in the lake and marshland regions (0.5 EPG).

Male study participants had a higher infection intensity than females, both among egg-positive persons (34 vs. 30 EPG) and the general population (0.4 vs. 0.3 EPG). No clear trend emerged when stratification was conducted by age, but intensity of infection was generally less variable between age groups in male participants and decreased somewhat with age in female participants. With regard to occupation, fisherman, boatmen, and preschool children had the highest geometric mean infection intensity (0.9 EPG), followed by farmers and students (0.4 EPG). The infection intensity generally decreased with higher educational level (Table).

Results of S. japonicum-related illness, as assessed by ultrasonography, and self-reported signs and symptoms, are shown in Appendix Table 2. Seropositive persons were more likely to report diarrhea over the past 2 weeks than seronegative persons. Degree of liver fibrosis was positively associated with a positive serologic result and with residency in the lake and marshland region. Persons living in the plain region with waterway networks had lower levels of liver fibrosis. No association was found between serologic status and hepatomegaly or splenomegaly.