Recreational water exposure and waterborne infections in a prospective salivary antibody study at a Lake Michigan beach

In a prospective observational study, seroconversion to a specific pathogen can serve as a marker of an incident infection, whether or not that infection is symptomatic or clinically diagnosed. While self-reported symptoms can be affected by reporting bias, seroconversion is likely to be free of this bias as it is based on objective measurements of antibody response. Non-invasive salivary antibody tests can be used instead of serum tests to detect seroconversions in prospective studies. In the present study, individuals and families were recruited at a Lake Michigan beach in Wisconsin in August 2011. Data on recreational water exposure and baseline saliva samples (S1) were collected at recruitment. Follow-up data on gastrointestinal symptoms were collected via a telephone interview approximately 10 days post-recruitment. Follow-up saliva samples were self-collected approximately 2 weeks (S2) and 30–40 days post-recruitment (S3) and mailed to the study laboratory. Samples were analyzed for immunoglobulin (Ig) G responses to recombinant antigens of three noroviruses and Cryptosporidium, as well as protein purification tags as internal controls, using an in-house multiplex suspension immunoassay on the Luminex platform. Responses were defined as ratios of antibody reactivities with a target protein and its purification tag. Seroconversions were defined as at least four-fold and three-fold increases in responses in S2 and S3 samples compared to S1, respectively. In addition, an S2 response had to be above the upper 90% one-sided prediction limit of a corresponding spline function of age. Among 872 study participants, there were seven (0.8%) individuals with seroconversions, including six individuals with seroconversions to noroviruses and two to Cryptosporidium (one individual seroconverted to both pathogens). Among 176 (20%) individuals who reported swallowing lake water, there were six (3.4%) seroconversions compared to one (0.14%) seroconversion among the remaining 696 individuals: the crude and age-standardized risk differences per 1000 beachgoers were 32.7 (95% confidence limits 5.7; 59.6) and 94.8 (4.6; 276), respectively. The age-adjusted odds ratio of seroconversion in those who swallowed water vs. all others was 49.5 (4.5; 549), p = 0.001. Individuals with a norovirus seroconversion were more likely to experience vomiting symptoms within 4 days of the index beach visit than non-converters with an odds ratio of 34 (3.4, 350), p = 0.003. This study contributed further evidence that recreational water exposure is associated with symptomatic and asymptomatic waterborne infections, and that salivary antibody assays can be used in epidemiological surveys of norovirus and Cryptosporidium infections.