Rotavirus disease is not nationally notifiable in the United States, yet surveillance of rotavirus infections is critical for identifying gaps in rotavirus vaccination coverage and understanding rotavirus epidemiology in the post-vaccination era. We used wastewater RNA concentrations of rotavirus collected over more than one year across the United States to infer rotavirus disease occurrence in the United States and compare occurrence patterns to clinical metrics of rotavirus infection and markers of rotavirus vaccination coverage. Rotavirus RNA was detected in wastewater samples throughout the study period in varying concentrations. Wastewater monitoring of rotavirus, coupled with the data analysis tools presented in this study, can improve public health in the USA by providing comprehensive insight into rotavirus occurrence and informing where to focus vaccination efforts.
Prior to our study, concentrations of rotavirus RNA in wastewater solids have only been reported in California. Rotavirus has been detected in wastewater sludge in several countries, but our study is the first to report concentrations in wastewater solids across the United States. Boehm et al. retrospectively measured rotavirus RNA concentrations in wastewater solids at two WWTPs in California between February 2021-April 2023 and observed median concentrations on the order of 10 gc/g. Concentrations were also orders of magnitude higher on a per mass basis compared to concentrations measured in matched wastewater liquid samples, justifying our decision to quantify rotavirus RNA in wastewater solids for this study. Herein, we observed median concentrations of rotavirus RNA in wastewater solids on the order of 10 gc/g based on samples from 185 WWTPs across 40 states and the District of Columbia between September 2023 and December 2024. Studies that measured rotavirus RNA concentrations in raw sewage observed seasonal occurrence patterns, as we observed in our study, with concentrations ranging on the order of 10-10 gc/L and 1-10 fluorescent foci/L.
Nationally, a rotavirus wastewater event occurred from 1 January to 23 June, which aligned with the winter-spring occurrence pattern of rotavirus infections according to clinical testing data. Wastewater measurements of rotavirus RNA were higher in fall 2023 (leading up to an even-numbered season) than in fall 2024 (leading up to an odd-numbered season) across the United States. We did not have longitudinal wastewater measurements from two consecutive winter-spring seasons, but our observations suggest a deviation from the typical biennial occurrence pattern of rotavirus. According to clinical testing data, odd-numbered seasons are generally associated with elevated incidence compared to even-numbered seasons in the post-vaccination era. The COVID-19 pandemic impacted the biennial pattern of rotavirus occurrence. Mathematical modeling suggests rotavirus transmission will not return to pre-pandemic equilibrium before 2030, which may explain why wastewater rotavirus RNA measurements seemingly did not align with the biennial occurrence pattern. Continued wastewater monitoring of rotavirus RNA can be a useful tool for empirically understanding how temporal patterns in rotavirus occurrence continue to develop.
The spatial progression of rotavirus occurrence in wastewater lacked national coherence. Elevated occurrence of rotavirus in wastewater began first in the South, followed by the Midwest, West, and Northeast. Prior to vaccines, annual rotavirus occurrence also lacked national coherence, beginning in the Southwest and ending in the Northeast. In the post-vaccination era, Burnett et al. suggest that areas with high birth rates and low vaccination coverage have the most rapid accumulation of susceptible individuals and consequently drive annual rotavirus occurrence. Indeed, rotavirus occurrence began in Oklahoma, Arkansas, and the western Gulf coast -- all areas in the South with high birth rate and low vaccination coverage -- according to clinical testing data across five seasons in the post-vaccination era. Wastewater monitoring of rotavirus RNA over consecutive seasons can be used alongside clinical testing data to characterize patterns in the spatial progression of rotavirus in the post-vaccination era.
Weekly wastewater measurements of rotavirus RNA were positively correlated with clinical metrics of rotavirus infections obtained from both NREVSS and Epic Cosmos at the national scale. Although the two clinical metrics themselves were positively correlated at the national scale (Fig. S12), the two datasets are distinct. Epic Cosmos is a research dataset that aggregates electronic health record data from 295 million patients. In contrast, NREVSS relies on voluntary reporting of rotavirus tests; fewer than 100 rotavirus test results were submitted to NREVSS each week during our analysis period. As shown in Fig. 2, test positivity from NREVSS was visually prone to more week-to-week variability compared to patient diagnoses from Epic Cosmos. Moreover, data can be accessed at subnational scales from Epic Cosmos, unlike from NREVSS. For example, we observed that wastewater measurements of rotavirus RNA were still positively correlated with clinical metrics of rotavirus infections at the state scale using data from Epic Cosmos (Fig. S13, Table S1). Nonetheless, electronic health record datasets like Epic Cosmos still suffer biases not present in wastewater monitoring data, such as bias towards individuals with symptomatic disease and access to healthcare. Rotavirus encounter diagnoses in Epic Cosmos mostly occur in the under-5-year age group (Fig. S14), suggesting infections in older populations are subclinical. Likewise, other studies observed a weak or lack of a positive correlation between rotavirus RNA concentrations in wastewater and traditional measures of disease surveillance, suggesting the presence of asymptomatic or subclinical infections in the contributing community. Wastewater monitoring represents a less-biased approach for inferring population-level disease occurrence compared to clinical or syndromic measures of disease surveillance.
Rotavirus wastewater event durations were generally shorter in sewersheds with markers of high rotavirus vaccination coverage, suggesting the utility of wastewater monitoring for identifying gaps in vaccination coverage. Exceptions include wastewater event durations not being associated with the proportion of Hispanic or Latino children or the pediatrician-to-child ratio, and not being highest in the Northeast. However, not all indirect markers of vaccination coverage identified by Ghaswalla et al. were directly available from publicly available sources. For example, we obtained public data on the pediatrician-to-child ratio, which describes access to pediatricians but not whether children actually seek care from pediatricians over family physicians. We also only conducted bivariate analyses between wastewater event durations and each indirect marker of rotavirus vaccination coverage. Future work might consider combining all markers of rotavirus vaccination coverage into a multivariable model with wastewater event duration as the outcome variable. Additionally, other metrics (e.g., peak wastewater measurement) may be important to consider as the outcome variable. Herein, we were interested in the effect of vaccination on prolonged rotavirus occurrence rather than an isolated outbreak, given that vaccines have been documented to shorten the rotavirus season. A limitation with our selected dependent variable, however, is that the wastewater event duration may be underestimated for some WWTPs and consequently some states. The wastewater sampling start and end dates differed across WWTPs (Table S2), so sampling may have ended in the midst of a wastewater event for some WWTPs. Nonetheless, the national wastewater event ended before 1 July 2024, which is when a large number of WWTPs ended sampling (mainly in California); it is likely that durations were underestimated for only a few WWTPs.
Prior to vaccines, ~2.7 million cases of rotavirus were reported each year in the United States, corresponding to about 1% of the population based on the population estimate from the 2000 census. Annual rotavirus incidence estimates in the United States are not available for the post-vaccination era but are expected to be substantially lower given the effectiveness of vaccines at preventing severe infection. Infected individuals may shed rotavirus RNA for consecutive days; therefore, modeled F values are likely more representative of disease prevalence rather than incidence. Thus, it is challenging to evaluate the accuracy of our model because rotavirus is monitored in terms of incidence (i.e., reported cases) rather than prevalence. Nonetheless, we expect the modeled F values in Fig. 5 to estimate higher rotavirus prevalence than could be determined from case reporting. The age-adjusted prevalence of asymptomatic rotavirus infection is 11% and rotavirus may be detected in about 20-30% of individuals without diarrhea. Rotavirus infection in adults also usually results in self-limiting symptoms, so adult infections are likely not captured in case reporting.
Fecal shedding dynamics are highly individualized, and the high sensitivity of our model to all input parameters demonstrates that precisely estimating disease occurrence from wastewater measurements is extremely challenging. Infected individuals who shed rotavirus RNA in feces in high concentrations or for long durations ("super shedders") likely contribute the most rotavirus RNA to wastewater. Future work is needed to better characterize fecal shedding dynamics and refine models so that disease occurrence may be estimated from wastewater measurements with less uncertainty. Additionally, rotavirus RNA in wastewater solids was measured using droplet digital reverse transcription-polymerase chain reaction (ddRT-PCR) without first subjecting samples to a "heat snap" (i.e., incubating samples at 99 °C for 5 min). Rotavirus is a double-stranded RNA (dsRNA), and the additional denaturing step may be important for quantifying RNA of dsRNA viruses using ddRT-PCR. Including a heat snap step can increase measured rotavirus RNA concentrations in wastewater, which would affect the modeled F reported herein.
Our study is the first to assess spatiotemporal occurrence patterns of rotavirus RNA in wastewater across the United States and compare occurrence patterns to markers of rotavirus vaccination coverage. Wastewater monitoring has been used previously to support polio eradication efforts. Wastewater monitoring of other vaccine-preventable diseases may similarly be valuable for identifying disparities in vaccination coverage and informing where vaccination campaigns should be targeted. Although this study focused on the United States, the global burden of rotavirus is disproportionately highest in Africa, Oceania, and South Asia and rotavirus vaccines have lower performance in countries with high child mortality. Therefore, wastewater monitoring for rotavirus to support vaccine prioritization could be especially impactful in low- and middle-income countries for reducing mortality. Reduced mortality could, in turn, boost economies by alleviating rotavirus-associated strain on healthcare systems. Continued work on adapting wastewater monitoring methods to decentralized or non-sewered settings is needed to scale wastewater monitoring of rotavirus globally.