Table 4. The observed and predicted proportion of the population with two consecutive S. japonicum infections.
Observed | Expected | Ratio (Obs./Exp.) | p-value‡ | |
Cohort 1, simple prediction model* | 5.07% | 3.41% | 1.48 | 0.00051 |
Cohort 1, exposure based prediction model† | 5.07% | 3.90% | 1.30 | 0.013 |
Cohort 2, simple prediction model* | 3.46% | 0.59% | 5.82 | 6.6×10−12 |
Cohort 2, exposure-based prediction model† | 3.46% | 1.68% | 2.06 | 0.00056 |
The expected prevalence of two consecutive infections was estimated based on the prevalence of infections at T1 and T2.
The expected prevalence of two consecutive infections was estimated accounting for S. japonicum exposure. The infection prediction model included water contact minutes by month and activity for all measures for which at least 20% of cohort participants reported exposure, age, sex, baseline village infection prevalence, county and year of infection test. Prediction models were fit separately for each cohort.
P-values were estimated assuming the number of individuals with two consecutive infections follows a binomial distribution, where is equal to the expected prevalence of two consecutive infections and is equal to the number of individuals in the full population. Thus the p-value is that of a two-sided, one-sample test assuming the probability of double-infections is equal to PDI.