Exploring trends in SARS-CoV-2 RNA concentrations in wastewater
In a recent study published on medRxiv* preprint server, researchers characterized trends in severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA levels in wastewater.
Wastewater-based epidemiology (WBE) relies on the levels of infectious disease markers in wastewater to assess the incidence of disease in the community. The coronavirus disease 2019 (COVID-19) pandemic has drawn attention to WBE. Specifically, SARS-CoV-2 RNA concentrations in sewage-decanted solids correlate well with incident clinical cases of COVID-19 (laboratory-confirmed cases) in the same sewer shed.
Additionally, trends in SARS-CoV-2 RNA concentrations in wastewater precede those in incident COVID-19 cases in communities. However, there is uncertainty in the interpretation and use of WBE data to aid in decision making. In addition, limited efforts have been made to actively monitor trends. The usual trend measures, such as rates of change, simple moving averages, etc., pay less attention to trend stability and statistical significance and can often be confusing/misleading.
About the study
In the current study, researchers used daily measurements of SARS-CoV-2 RNA in wastewater to compare three trend analysis methods and assess their performance. Daily sampling of the San José-Santa Clara Regional Wastewater Treatment Plant began November 15, 2020 through September 15, 2022. The authors used data on SARS-nucleocapsid (N) gene concentrations. CoV-2 as gene copies/g dry weight normalized by mild pepper mottle virus (PMMoV) concentrations.
Three analytical parameters – percent change (PC), Mann-Kendall (MK) trend test, and relative strength index (RSI) were used to identify trends in N or PMMoV over time. The RSI is useful in informing trend stability, while the other two inform statistical significance. The RSI was calculated for the seven-day right-aligned moving average of N or PMMoV using a 14-day look-back period.
For PC, the authors used the formula that the US Centers for Disease Control and Prevention (CDC) uses to calculate trends in SARS-CoV-2 RNA concentrations in wastewater. The MK trend test, a nonparametric test, was used to assess monotonic trends in a time series data set. The MK trend test was applied to test raw N or PMMoV data using a look-back period of 14 days.
Heat maps were created for the entire analysis period and separately for three waves of COVID-19 caused by 1) the SARS-CoV-2 delta, 2) Omicron BA.1 and 3) Omicron BA.2 variants and BA.5. Additionally, the authors downsampled the dataset to repeat the PC and MK trend test methods. The downsampled data set was generated for all combinations covering frequencies between two samples/week and six samples/week.
SARS-CoV-2 RNA trends in wastewater were stratified as increasing, decreasing, or stable using the three parameters. In addition, the MK trend test and PC calculated upside and downside sensitivity and specificity for each subsampled data set. Up/down sensitivity was defined as the correct identification of an up/down trend, while specificity was the ability to identify no trend.
All methods identified increasing, decreasing and stable trends in SARS-CoV-2 RNA concentrations in wastewater. The MK trend test and the PC identified uptrends earlier than RSI at the start of the Delta and BA.2/BA.5 waves. In particular, the MK trend test identified the uptrend 17 and 12 days before RSI at the start of the Delta and BA.2/BA.5 waves, respectively.
Similarly, PC identified the uptrend 16 and 26 days before RSI at the start of the Delta and BA.2/BA.5 waves, respectively. Additionally, the MK trend test and the PC identified downtrends at the end of the COVID-19 waves before RSI. In subsampling analyses, the MK trend test and PC had low sensitivity (upward/downward) for low sampling frequencies, but improved with increasing frequency.
The MK trend test achieved acceptable upward/ downward sensitivity with a minimum of five samples/week. On the other hand, PC achieved acceptable top-down/top-down sensitivity with at least four samples/week. Notably, the specificity remained similar and very high for both methods at all sampling frequencies.
Both the MK trend test and the PC provided more early warning of uptrends and downtrends using daily data than the RSI. Subsampling analyzes suggested that a sampling frequency of at least four and five samples per week was necessary to identify trends using the PC and MK trend test, respectively. Since the PC required fewer samples/week than the MK trend test, WBE programs with budget constraints may prefer the PC.
Overall, the MK trend test and the PC are inference-based methods and can be used to classify trends in a standard way. These trend analysis approaches can be adopted by WBE programs to inform public health departments of changing trends in COVID-19 cases, particularly as the rate of clinical testing for SARS-CoV-2 decreases.
medRxiv publishes preliminary scientific reports that are not peer-reviewed and, therefore, should not be considered conclusive, guide clinical practice/health-related behaviors, or treated as established information.