Comparability of 24-hour composite and grab samples for detection of SARS-2-CoV RNA in wastewater

Abstract Wastewater surveillance is a cost-effective way to monitor pathogen prevalence and transmission patterns in the entire community. Here, we compare 24-hour composite and grab samples collected during September 2020 from several municipalities in New York State to detect SARS-CoV-2. A total of 45 paired samples (90 total samples) from three counties and 14 wastewater treatment plants were available for analysis. The categorical comparison (SARS-CoV-2 genetic material detected and quantifiable, genetic material detected but below the limits of quantification, and genetic material not detected) between the grab and composite samples was quite strong, with 91.1% agreement (kappa P-value < .001). The correlations among the quantifiable grab and composite samples were statistically significant yet modest for SARS2-CoV RNA (Pearson correlation = 0.44, P = .02), crAssphage cDNA (Pearson correlation = 0.36, P = .02), and crAssphage DNA (Pearson correlation = 0.46, P = .002). We found good comparison between grab and 24-hour composite samples for detecting SARS-CoV-2 RNA from municipal wastewater treatment plants. Grab sampling is an efficient and cost-effective method to monitor for the presence of SARS-CoV-2 in the entire community.


Introduction
Waste water surv eillance for infectious diseases has been used for decades, first gaining popularity during polio eradication campaigns (Vaccines and Biologicals World Health Organization 2003 ).Waste water surv eillance is a cost-effective way to monitor pathogen pr e v alence and tr ansmission patterns in the entire community (Larsen and Wigginton 2020 ).With the start of the COVID-19 pandemic, se v er al gr oups , including ours , were able to demonstrate the feasibility of detecting SARS-CoV-2 RNA in wastewater (Medema et al . 2020, Peccia et al . 2020, Wurtzer et al . 2020, Wilder et al . 2021 ).Soon after, municipalities across the globe decided to use the science to monitor SARS-CoV-2 transmission in their communities (Naughton et al . 2021 ).Ho w e v er, this unpr ecedented interest caused availability shortages in w astew ater sampling equipment.
Gener all y, continuous flow-pr oportional sampling, wher e a small sample of w astew ater is collected at a defined volume interv al, is consider ed the gold standar d.Ho w e v er, this is not al ways possible due to the inability to install flow-measuring equipment at sampling locations .T her efor e , 24-composite samples , where a small sample of w astew ater is collected at regular time intervals (such as every 15 minutes) for 24 hours and then combined together are often used as an industry standard (Schaeffer et al . 1980, Brumelle et al . 1984, Cornman et al . 2018, Water Research Foundation T 2020 ).This sampling method can account for variations in w astew ater c har acteristics fr om the population it is mon-itoring and increase the validity of the sample.Ho w ever, the autosamplers r equir ed to collect this type of sample ar e expensiv e, subject to availability issues, and cannot be used at all collection points.When an autosampler is not a vailable , a one-time grab sample is often used instead.This method, ho w e v er, is highl y subjected to w astew ater discharge patterns and the timing of the grab sample should be chosen strategically.Water soluble pharmaceuticals, personal car e pr oducts, and illicit drugs are particularly susceptible to inadequate sampling strategies, and it is difficult to determine if variations in results are due to real changes or due to the sampling method (Ort et al . 2010 ).The SARS-CoV-2 is a solids-associated RN A virus; w e aimed to assess the utility of grab samples as a tool to detect the virus.
During the summer of 2020, New York State implemented a pilot pr ogr am in se v er al m unicipalities to monitor SARS-CoV-2 tr ansmission in waste water.During the course of this pr ogr am, we collected both a 24-hour composite and a grab sample from sever al m unicipalities.Her e, we describe the comparability of these methods for detection of SARS-CoV-2 and crAssphage, a human fecal indicator (Stachler et al . 2017 ).

Study design
The New York State Department of Health commissioned a pilot study of w astew ater surveillance for SARS-CoV-2 in August of 2020.As man y differ ent SARS-CoV-2 anal ytical pr ocedur es hav e been de v eloped by a host of public and priv ate entities, we designed a study to compare results from 24-hour composite samples to well-timed grab samples, collected during the morning diurnal peak.This comparison was part of a greater quality assessment pr ogr am to determine sampling and anal ytical consistency, re peatability, effecti ve range, and limit of detection.

Grab and 24-hour composite sample collection
Both grab and 24-hour composite samples were collected from specified locations in the cities of Albany, Buffalo, and Newburg in New York State as shown in Table 1 .Autosamplers were set to collect at a frequency and sampling volume within parameters set forth by the US Environmental Protection Agency.A re presentati ve sample was collected from the 24-hour composite sampler, the samples were iced or refrigerated during the collection process.
A grab sample was obtained at the same location and collected at the end of the 24-hour collection period.Both samples were transported in a cooler with ice to maintain a 4 • C temper atur e.
Both samples arrived at the anal ytical labor atory within 24 hours of collection.Sample collection follo w ed standar dized oper ating pr ocedur es (SOPs) for sampling and a quality assurance/quality control (QA/QC) plan for all routine and point monitoring sampling.This plan included the use of field blank samples, blind duplicate samples, and matrix spike duplicate samples.The use of the SOPs minimized the variability associated with sampling and sample splitting in order to e v aluate the inher ent v ariability associated with w astew ater, as w ell as the labor atory v ariability.

SARS-2-CoV RNA and crAssphage DNA and RNA quantification
SARS-CoV-2 RNA and crAssphage DNA and cDNA were quantified as pr e viousl y described (Wilder et al . 2021 ).In brief, samples wer e pr ocessed using the ultr acentrifugation thr ough a sucr ose cushion tec hnique befor e nucleic acid extr action and PCR quantification.This method has a limit of quantification of five copies per ml (Wilder et al . 2021 ).All w astew ater samples w er e anal yzed in triplicate.

Case data collection
In order to confirm that cases were present in the sampled sewer sheds, we matched COVID-19 diagnosed cases and diagnostic tests from the New York State Electronic Clinical Laboratory System (ECLRS; New York State Department of Health 2021 ) to sewershed pol ygons gener ated in consultation with m unicipal sanitation engineers using the home address of each COVID-19 case.

Sta tistical anal ysis
Eac h waste water sample could have one of the three possible SARS-CoV-2 qualitative outcomes: genetic material detected and quantifiable, genetic material detected but below the limits of quantification, or genetic material not detected.The % a gr eement and kappa statistic were calculated for the categorical outcomes of the paired samples .T he amount of genetic material in the quantifiable samples were log transformed.Among the log transformed quantifiable samples, the correlations between the grab and composite samples were plotted, pairwise Pearson corr elation coefficients wer e calculated, and quantile-quantile plots wer e gener ated.Equality of v ariances between the gr ab and composite samples were compared using Levene's Robust Test statistic.We assessed four separate measures: log-transformed SARS-CoV -2, log-transformed SARS-CoV -2 normalized by dividing logtr ansformed crAsspha ge DNA, log-tr ansformed crAsspha ge DNA, and log-transformed crAssphage cDN A. Tw o-sided P -values less than .05were considered significant.These analyses were completed in Stata version 16.0.

Results
A total of 45 paired samples (90 total samples) from three wastewater treatment plant sewer sheds associated with 14 w astew ater sampling locations in Upstate New York were included in this analysis (Table 1 ).All samples were collected in September of 2020.A total of 28 samples did not have detectable SARS-CoV-2 RN A, one w as detectable but not quantifiable, and 61 had quantifiable SARS-CoV-2 RNA.These r esults wer e detected during a time of r elativ el y low SARS-CoV-2 tr ansmission, with the waste water catc hment ar eas r eporting less than fiv e cases per day, or less than 15 cases per 100 000 population, on av er a ge (Table 1 ).All of the samples had quantifiable crAssphage DNA and 88 samples had quantifiable crAssphage cDNA.The remaining two samples were nondetectable for crAssphage cDNA.
The categorical comparison between the grab and composite samples sho w ed 91.1% a gr eement (ka ppa P -v alue < .001)for detecting SARS-CoV-2 RNA (Table 2 ).There was no difference in the v ariance between gr ab and composite samples for the SARS-CoV-2 to crAssphage ratio ( P = .23),the SARS-2-CoV RNA ( P = .36),and crAssphage DNA ( P = .43).Grab samples had lo w er variance than composite samples when assessing crAssphage cDNA ( P = .04).The quantile-quantile plots also support that the grab and composite samples are similar, except for the crAssphage cDNA (Fig. 1 ).The correlations among the quantifiable grab and composite samples were statistically significant yet modest for SARS-CoV-2 RN A (P earson correlation = 0.44, P = .02),crAssphage cDNA (Pearson correlation = 0.36, P = .02),and crAssphage DN A (P ear-F igure 2. Correlations betw gr ab and composite samples (log v alues).son correlation = 0.46, P = .002;Fig. 2 ).Ho w ever, there w as a wide range of variation in the correlation coefficients by county, with most not statisticall y significant, likel y attributable to the small number of paired samples available for analysis in each county (Table 3 ).

Discussion
We found good comparison between grab and 24-hour composite samples for detecting SARS-CoV-2 RNA fr om m unicipal waste water treatment plants as well as for crAssphage DNA.Our results suggest that grab samples may be sufficient to detect SARS-CoV-2 RNA in order to monitor presence/absence o vertime .Ho w ever, there is less agreement between the grab and composite samples when quantifying the amount of SARS-CoV-2 RNA in wastewater.Ther efor e, the a ppr opriate sampling str ategy depends on the goals of the surveillance program.
A comparison of samples collected from manholes during a period of high transmission in Iran found that grab and composite samples gener all y a gr eed for detecting SARS-CoV-2 RN A. Ho w e v er, gr ab samples r eported less SARS-CoV-2 RNA in the w astew ater than the composite samples (Rafiee et al . 2021 ).The lar ge dail y variations in the amount of human fecal matter at the subsewer shed le v el likel y contributed to the poor er performance of the gr ab samples in this study (Rafiee et al . 2021 ).A gr oup in Vir ginia found good a gr eement between gr ab and composite samples during a period of low transmission from samples collected at a municipal w astew ater treatment facility (Curtis et al . 2020 ).Ho w e v er, neither of these studies examined the comparability of the types of sam-pling methods for detecting crAssphage DNA or other markers of human fecal matter.
This study was completed during a time of r elativ el y low COVID incidence at municipal w astew ater treatment plants.It is unclear how the grab samples would compare to the 24-hour composite samples during periods of more intense transmission.Our analysis demonstrates that grab and 24-hour composite samples perform equally well for detecting (or not detecting) SARS-CoV-2 RNA in m unicipal waste water.Ther efor e, gr ab samples ar e an efficient and cost-effective method to monitor for the presence of SARS-CoV-2 in the entire community.This methodology can be used in place of autosamplers at locations where autosamplers are not feasible , in resource-limited settings , and when there are supply chain disruptions.

Ac kno wledgments
This work was made possible by the participating wastewater treatment plants and their operators .T he NY state pilot study was conducted in collaboration with Arcadis and Quadrant Biosciences.Funding: this work was funded by the participating municipal w astew ater treatment plants, a w ork or der fr om the Ne w York State Department of Health and the New York State Division of Environmental Conservation to conduct a pilot program, and seed grants from the Syracuse University and SUNY ESF.T.Z.acknowledges the support from the Faculty Fellows Program administered by the Syracuse Center of Excellence for Environmental and Energy Systems (SyracuseCoE) through an aw ar d from the New York State Department of Economic De v elopment under aw ar d number #C150183.

Table 1 .
Basic c har acteristics for gr ab and composite sampling locations in Ne w York State, 2020.

Waste wa ter treatment plant Sampling point Number of paired samples Description Population * Mean daily COVID cases during September 2020 Mean daily COVID cases during September 2020 per 100 000 population
* Estimated total.

Table 2 .
Detection a gr eement among gr ab and composite samples for SARS-2-Cov RNA, Upstate NY, 2020.
* Log values used in correlation analysis.