High frequency, high throughput quantification of SARS-CoV-2 RNA in wastewater settled solids at eight publicly owned treatment works in Northern California shows strong association with COVID-19 incidence

A number of recent retrospective studies have demonstrated that SARS-CoV-2 RNA concentrations in wastewater are associated with COVID-19 cases in the corresponding sewersheds. Implementing high-resolution, prospective efforts across multiple plants depends on sensitive measurements that are representative of COVID-19 cases, scalable for high throughput analysis, and comparable across laboratories. We conducted a prospective study across eight publicly owned treatment works (POTWs). A focus on SARS-CoV-2 RNA in solids enabled us to scale-up our measurements with a commercial lab partner. Samples were collected daily and results were posted to a website within 24-hours. SARS-CoV-2 RNA in daily samples correlated to incidence COVID-19 cases in the sewersheds; a 1 log10 increase in SARS-CoV-2 RNA in settled solids corresponds to a 0.58 log10 (4X) increase in sewershed incidence rate. SARS-CoV-2 RNA signals measured with the commercial laboratory partner were comparable across plants and to measurements conducted in a university laboratory when normalized by pepper mild mottle virus PMMoV RNA. Results suggest that SARS-CoV-2 RNA should be detectable in settled solids for COVID-19 incidence rates > 1/100,000 (range 0.8 - 2.3 cases per 100,000). These sensitive, representative, scalable, and comparable methods will be valuable for future efforts to scale-up wastewater-based epidemiology.


Abstract 23
A number of recent retrospective studies have demonstrated that SARS-CoV-2 RNA 24 concentrations in wastewater are associated with COVID-19 cases in the corresponding 25 sewersheds. Implementing high-resolution, prospective efforts across multiple plants depends 26 on sensitive measurements that are representative of COVID-19 cases, scalable for high 27 throughput analysis, and comparable across laboratories. We conducted a prospective study 28 across eight publicly owned treatment works (POTWs). A focus on SARS-CoV-2 RNA in solids 29 enabled us to scale-up our measurements with a commercial lab partner. Samples were 30 collected daily and results were posted to a website within 24-hours. SARS-CoV-2 RNA in daily 31 samples correlated to incidence COVID-19 cases in the sewersheds; a 1 log10 increase in 32 SARS-CoV-2 RNA in settled solids corresponds to a 0.58 log10 (4X) increase in sewershed 33 incidence rate. SARS-CoV-2 RNA signals measured with the commercial laboratory partner 34 were comparable across plants and to measurements conducted in a university laboratory when 35 normalized by pepper mild mottle virus PMMoV RNA. Results suggest that SARS-CoV-2 RNA 36 should be detectable in settled solids for COVID-19 incidence rates > 1/100,000 (range 0.8 -37 2.3 cases per 100,000). These sensitive, representative, scalable, and comparable methods will 38 be valuable for future efforts to scale-up wastewater-based epidemiology. 39 40 Importance 41 Access to reliable, rapid monitoring data is critical to guide response to an infectious disease 42 outbreak. For pathogens that are shed in feces or urine, monitoring wastewater can provide a 43 cost-effective snapshot of transmission in an entire community via a single sample. In order for 44 a method to be useful for ongoing COVID-19 monitoring, it should be sensitive for detection of 45 low concentrations of SARS-CoV-2, representative of incidence rates in the community, 46 scalable to generate data quickly, and comparable across laboratories. This paper presents a 47 method utilizing wastewater solids to meet these goals, producing measurements of SARS-48 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 Introduction 54 55 The COVID-19 pandemic has prompted the rapid and widespread maturation of wastewater 56 based epidemiology (WBE). Through the first year of the pandemic, retrospective analyses of 57 wastewater samples for SARS-CoV-2 RNA demonstrated strong correlations between SARS-58 CoV-2 target concentrations and infection incidence 1-8 . Wastewater monitoring can therefore 59 provide a cost-effective snapshot of transmission in a community by using just one sample to 60 provide information that is unbiased by access to testing or symptom status. To be of most 61 value for public health officials, the methods used to produce data should be 1) sensitive to 62 detect low levels of viral RNA in wastewater, 2) representative of the disease rates in the 63 community, 3) scalable to provide high-throughput results with a short turnaround time, and 4) 64 comparable between labs and different approaches. 65 66 Numerous independent COVID-19 WBE efforts have covered monitoring at a range of scales, 67 from the building level to country-wide implementation 7,9,10 . The methods employed have also 68 varied, with the vast majority focusing on the liquid fraction of municipal wastewater. Due to the 69 dilute nature of SARS-CoV-2 RNA in wastewater, most methods that focus on influent require a 70 number of preanalytical steps to concentrate SARS-CoV-2 prior to extracting the viral RNA. 71 Depending on the method, these steps include ultrafiltration 2,5 , organic flocculation coupled with 72 centrifugation 3,11 , and charged membrane filtration 12 . Each of these listed concentration steps 73 add significant time, equipment, and personnel requirements to SARS-CoV-2 RNA 74 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 20, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 quantification in wastewater. For effective monitoring, methods need to be scalable. SARS-75 CoV-2 RNA measurements should be prospective as opposed to retrospective, should provide 76 high-resolution (e.g. daily) data on SARS-CoV-2 levels, and should result in data within hours or 77 days of sample collection. As wastewater monitoring expands to more communities, methods 78 focused on quantifying SARS-CoV-2 in liquid are not readily amenable to scale-up and 79 automation due to the additional steps needed to prepare the samples. 80 The majority of SARS-CoV-2 RNA in wastewater originates in the feces of infected individuals. 81 Even after mixing with wastewater liquid, coronaviruses have a stronger affinity to the solid 82 fraction of wastewater, even higher than the affinities of nonenveloped viruses 13 . Studies 83 comparing SARS-CoV-2 RNA concentrations in the liquid and solid fractions of real wastewater 84 have demonstrated that the solids harbor 3-4 orders of magnitude higher concentrations of 85 SARS-CoV-2 RNA on a per mass basis than liquid influent 3,14 . As a result, the settling of 86 wastewater solids that happens during primary treatment at most wastewater treatment plants 87 serves as a built-in concentration step for wastewater monitoring. SARS-CoV-2 RNA 88 quantification from these samples simultaneously requires less preanalytical processing and, on 89 a per mass basis, will result in higher measured concentrations than measurements made on 90 the liquid fraction of wastewater. If plants do not have a primary clarifier or samples are taken at 91 the sub-sewershed level, solids may still be concentrated from influent using standard methods 92 15 . SARS-CoV-2 monitoring focused on settled solids may therefore be both more sensitive and 93 more conducive to scale-up and automation. 94 In this study, we demonstrate that a SARS-CoV-2 monitoring project focused on wastewater 95 solids can be readily scaled to produce sensitive results that are representative of COVID-19 96 incidence through a high-frequency effort conducted through a commercial laboratory with <24 97 hour turn-around times. We initiated a prospective monitoring project across eight POTWs in the 98 greater San Francisco Bay and Sacramento areas of California to measure SARS-CoV-2 RNA 99 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 20, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 5 in daily samples. Results were consistently reported to stakeholders and agencies within 24 100 hours of sample collection. Here we show that SARS-CoV-2 RNA in daily samples correlates 101 strongly to COVID-19 incidence rates in the sewersheds. Results indicate SARS-CoV-2 RNA 102 signals are comparable across plants and across laboratories using different measurement 103 methods. We present the empirical sensitivity of the measurements to incidence rates in the 104 sewersheds. POTWs serve between 66,000 and 1,500,000 residents and have permitted flows between 8.5 120 and 181 million gallons per day (Table 1). 121 122

Sample Collection 123
Samples were collected by POTW staff using sterile technique in clean, labeled bottles 124 provided by our team. POTWs were not provided additional compensation for participation. 125 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021.  (Table  130 1). At Gil, solids were settled from a 24 h composite influent sample using standard method 131 160.5 15 . Samples were immediately stored at 4°C and transported to the lab where processing 132 began immediately (within 6 hours of collection). 133

Sample Preparation 135
The solids were dewatered by centrifugation at 24,000 x g for 30 minutes at 4°C. The 136 supernatant was aspirated and discarded. A 0.5 -1 g aliquot of the dewatered solids was dried 137 at 110°C for 19-24 hrs to determine its dry weight. Bovine coronavirus (BCoV) was used as a 138 positive recovery control. Each day, attenuated bovine coronavirus vaccine (PBS Animal Health, 139 Calf-Guard Cattle Vaccine) was spiked into DNA/RNA shield solution (Zymo Research) at a 140 concentration of 1.5 µL /mL. Dewatered solids were resuspended in the BCoV-spiked DNA/RNA 141 shield to a concentration of 75 mg/mL. This concentration of solids was chosen as previous 142 work titrated solutions with varying concentrations of solids to identify a concentration at which 143 inhibition of the SARS-CoV-2 assays was minimized (data not shown). Between five and ten 144 5/32" Stainless Steel Grinding Balls (OPS Diagnostics) were added to each sample which was 145 subsequently homogenized by shaking with a Geno/Grinder 2010 (Spex SamplePrep). Samples 146 were then briefly centrifuged to remove air bubbles introduced during the homogenization 147 process, and then vortexed to re-mix the sample. 148 149

RNA Extraction 150
. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. OneStep-96 PCR Inhibitor Removal Kit. Extraction negative controls (water) and extraction 154 positive controls were extracted using the same protocol as the homogenized samples. The 155 positive controls consisted of 500 copies of SARS-CoV-2 genomic RNA (ATCC® VR-1986D™) 156 in the BCoV-spiked DNA/RNA shield solution described above. 157 158 Droplet Digital PCR 159 RNA extracts were used as template in digital droplet RT-PCR assays for SARS-CoV-2 N, S, 160 and ORF1a RNA gene targets in a triplex assay, and BCoV and PMMoV in a duplex assay (see 161   Table S1 for primer and probe sequences, purchased from IDT). PMMoV is highly abundant in 162 human stool and domestic wastewater globally 19,20 and is used here as an internal recovery 163 and fecal strength control. Undiluted extract was used for the SARS-CoV-2 assay template and 164 a 1:100 dilution of the extract was used for the BCoV / PMMoV assay template. Digital RT-PCR 165 was performed on 20 µl samples from a 22 µl reaction volume, prepared using 5.5 µl template, 166 mixed with 5.5 µl of One-Step RT-ddPCR Advanced Kit for Probes (Bio-Rad 1863021), 2.2 µl 167 Reverse Transcriptase, 1.1 µl DTT and primers and probes at a final concentration of 900 nM 168 and 250 nM respectively. Droplets were generated using the AutoDG Automated Droplet 169 Generator (Bio-Rad). PCR was performed using Mastercycler Pro with the following protocol: 170 reverse transcription at 50°C for 60 minutes, enzyme activation at 95°C for 5 minutes, 40 cycles 171 of denaturation at 95°C for 30 seconds and annealing and extension at either 59°C (for SARS-172 CoV-2 assay) or 56°C (for PMMoV/BCoV duplex assay) for 30 seconds, enzyme deactivation at 173 98°C for 10 minutes then an indefinite hold at 4°C. The ramp rate for temperature changes were 174 set to 2°C/second and the final hold at 4°C was performed for a minimum of 30 minutes to allow 175 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ; https://doi.org/10.1101/2021.07.16.21260627 doi: medRxiv preprint 8 the droplets to stabilize. Droplets were analyzed using the QX200 Droplet Reader (Bio-Rad). All 176 liquid transfers were performed using the Agilent Bravo (Agilent Technologies). 177 178 Each sample was run in 10 replicate wells, extraction negative controls were run in 7 wells, and 179 extraction positive controls in 1 well. In addition, PCR positive controls for SARS-CoV-2 RNA, 180 BCoV, and PMMoV were run in 1 well, and NTC were run in 7 wells. Positive controls consisted 181 of BCoV and PMMoV gene block controls (dsDNA purchased from IDT) and gRNA of SARS-182 CoV-2 (ATCC® VR-1986D™). Results from replicate wells were merged for analysis. where the recovery of BCoV was less than 1%. 196 197 Ancillary wastewater data. Wastestream influent total suspended solids (TSS) concentrations in 198 mg/L data were obtained from POTW staff. If TSS measurements were not coincident on the 199 day that a sample was taken, the TSS value for that day was estimated using linear 200 interpolation. 201 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. for each sewershed were obtained from local or state sources through data-use agreements. 205 Case data were aggregated within the sewersheds based on georeferenced reported home 206 addresses, which were delineated using the POTW-specific GIS shape files. COVID-19 207 incidence rates per 100,000 population were calculated using the estimated population served 208 in each sewershed. The latter is justified owing to the "weekend effect" associated with a reduction in test seeking 216 behavior, testing availability, and result reporting 22 . Hereafter, any reference to incidence or 217 incidence rate will refer to the value from the 7-day smoothed average. 218 219 Incidence rates were compared to SARS-CoV-2 RNA concentrations, SARS-CoV-2 RNA 220 concentrations normalized by PMMoV concentrations (CPMMoV), and SARS-CoV-2 RNA 221 concentrations scaled by the factor F= Kdp(1+KdTSS)/(CPMMoVKd(1+KdpTSS)) where Kd and Kdp 222 are the partitioning coefficients for SARS-CoV-2 and PMMoV, respectively, and the other terms 223 have been defined. The scaling factor falls from a mass balance model relating the number of 224 SARS-CoV-2 fecal shedders in a sewershed to the concentration of SARS-CoV-2 RNA in 225 settled solids 7 and will hereafter be referred to as F for simplicity. In applying F to the data, we 226 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ; https://doi.org/10.1101/2021.07.16.21260627 doi: medRxiv preprint used Kd = 1000 and Kdp = 100 7 . Results were similar when the analysis was repeated with 227 varying Kd and Kdp between 100 and 10 4 ml/g (data not shown). 228 229 Non-parametric Kendall's tau and Kruskal-Wallis methods were used to test hypotheses 230 regarding associations and trends as data were neither normally nor log-normally distributed 231 based on Shapiro-Wilk tests. Linear regressions were used to assess slopes describing 232 relationships between incidence rates and N gene concentrations normalized by PMMoV for 233 each POTW and for all POTWs aggregated. To account for variability of wastewater 234 measurements, Kendall's tau empirical p-values and regression coefficients m were determined 235 using 1,000 bootstrap re-samplings that incorporate non-detect measurements and 236 measurement errors 3 , and median tau, regression coefficients m, standard error, R 2 , and 237 empirical p are reported. For trimmed data, bootstrapping utilized 95% confidence intervals 238 accounting for variability in the 5 measurements included in the trimmed average. 239

240
The minimum incidence rate at which wastewater solids contain measurable SARS-CoV-2 RNA 241 was estimated for each POTW. Predictions were calculated within the bootstrapping approach 242 by using coefficients from the observed linear relationship between log10-transformed COVID-19 243 incidence rate and log10 N cp/g measured in wastewater solids for each bootstrapped sample to 244 predict the incidence rate when wastewater solids contained 750 cp/g of the N gene. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ; https://doi.org/10.1101/2021.07.16.21260627 doi: medRxiv preprint 11 rates. These data were used to assess whether trends observed with data in the present study 253 are distinct from those reported by Wolfe et al. 7 . The empirical regression coefficients for the 254 two data sets were compared using Kruskal Wallis test to test the null hypothesis of no 255 difference between the distribution of coefficients. Those authors measured N1 and PMMoV 256 using a different workflow than used in the present study. The detection limit of N1 reported by 257 those authors was ~40 cp/g; in assessing associations using these data, the bootstrapping 258 approach samples from a uniform distribution defined by 0 and 40 cp/g to assign a value to non-259 detects in the data set. respectively. BCoV recoveries were, on average 57% (standard deviation = 39%) and all were 265 above 1% ( Figure S1). Sample standard deviations for the SARS-CoV-2, PMMoV, and BCoV 266 recovery quantification estimated from the merged wells were, on average 19%, 19%, and 14% 267 of the measurement. As the samples were extracted ten times and each extract analyzed in one 268 of 10 replicate wells which were merged, the replicate variability incorporates variation from both 269 RNA extraction and RT-ddPCR with a heterogeneous solids sample. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. Measurement overview. Across all samples, PMMoV ranged from 4.3 x 10 7 to 7.1 x 10 9 cp/g 280 (average = 6.6 x 10 8 ). PMMoV was different between POTW (Kruskal-Wallis p<10 -15 ) with 281 Ocean tending to have the lowest PMMoV and Gil the highest ( Figure S2). SARS-CoV-2 RNA 282 gene concentrations ranged from 630 to 3.7x10 6 (N gene), ND to 3.2x10 6 (S gene) and ND to 283 3.0x10 6 (ORF1a) cp/g across all samples. A total of 4 samples returned non-detects during this 284 period (1 for S and 3 for ORF1a) and the value of 300 cp/g (approx half the detection limit) was 285 substituted for these for further analysis. N, S, and ORF1a gene concentrations were highly 286 correlated (rp ranged from 0.97 to 0.98 for log10-transformed data aggregated across plants). 287 Pairwise linear regressions between log10-transformed N, S, and ORF1a concentrations 288 returned slopes ~1 ( Figure S3). Results with untransformed variables were similar (see SI). 289 Therefore, further analyses in this paper focuses on the N gene alone. 290

Relationship between SARS-CoV-2 RNA in solids and incident cases. Concentrations of SARS-291
CoV-2 RNA rose over the "winter surge" in COVID-19 incidence in November and December of 292 2020 and declined to lower levels at the end of March (Figure 1). There is an apparent dip in 293 incidence rate at the peak of the surge; this was probably due to decreased test seeking 294 behavior and reduced testing availability during the week between Christmas and New Year's 295 Day. During the time period of this study there were no days on which any POTW achieved a 296 non-detect across the three genes, nor were there days where incident case numbers in the 297 sewershed were 0. Over the duration of the study, the lowest smoothed incident rates observed 298 across the eight POTWs ranged from 1.9 to 8.5 cases per 100,000 people (at Dav and Sac, 299 respectively). These correspond to the low numbers of daily smoothed incident cases ranging 300 from 1.3 to 125.7 in these two sewersheds. Wastewater data availability preceded case data 301 availability; case reporting delays in this region at the time were greater than 3 days. 302 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021.  Figure S4-S6). Within POTW, associations 305 were significantly enhanced when N was normalized by PMMoV, or scaled by F at 5 of the 8 306 POTW, although the effect size was small (effect size < 0.1, Kruskal Wallis p<0.001); at Gil, 307 Sun, and SVCW, the association was significantly weakened (effect size < 0.1, Kruskal Wallis 308 p<0.001). When 7-d trimmed average N gene concentrations were used in lieu of raw N gene 309 concentrations the significance and direction of differences in association between N, N 310 normalized by PMMoV and N scaled by F and incidence rate were unchanged at each plant. 311 Kendall's tau was significantly higher for each of these measurements when data was trimmed, 312 although the effect size was small (effect size <0.1, Kruskal Wallis all p<10 -15 ). 313 314 When data from the eight POTWs are aggregated, a strong association between wastewater 315 concentrations of SARS-CoV-2 RNA and incidence rate persists (Figure 2). Kendall's tau 316 between raw N concentrations and incidence rates aggregated across plants is 0.66 (p<0.001); 317 tau decreases to 0.58 (p<10 -15 ) when data are normalized by PMMoV or scaled by F (Table 2, 318 Figure S7). Associations between incidence rate and 7-d trimmed average wastewater data are 319 similar to those between incidence rate and raw wastewater data (Table 2). Linear regressions 320 between COVID-19 incidence rate and wastewater values suggest that for a 1 log10 increase in 321 N cp/g, there is a 0.59 (± 0.01 standard error) log10 increase in incidence rate (R 2 = 0.67), and 322 for N normalized by PMMoV and N scaled by F there is a 0.58 (±0.02) log10 increase in COVID-323 19 incidence (R 2 = 0.58). were added to the aggregated plots ( Figure 3). Those data were generated in a different 327 laboratory using a different method from those used for the data reported herein, and represent 328 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 data from diverse POTWs in California, Illinois, and New York during a different phase of the 329 pandemic (Spring-Fall 2020). When displayed as raw N gene concentrations versus incidence 330 rate, the Wolfe et al. data do not fall on the same data cloud as the data generated in the 331 present study and the slope of the linear regression lines through each data set are divergent 332 (slope = 0.13 ± 0.03 and 0.59 ± 0.01 log10 incidence rate/log10 N cp/g for Wolfe et al. and the 333 data from this paper, respectively). However, when all the data are normalized by PMMoV or 334 scaled by a factor that includes TSS and partitioning coefficients, the Wolfe et al. data collapse 335 onto the data generated in this study. After normalizing by PMMoV or scaling by F, the 336 regression lines for the two methods remain significantly different (Kruskal Wallis p<0.001 for 337 all), however the slopes describing the relationship between wastewater values and incidence 338 rate are similar between the two methods after applying these approaches (slope = 0.24 ± 0.03 339 and 0.58 ± 0.02 log10 incidence rate/log10 wastewater value for both normalized and scaled data 340 from Wolfe et al. and this paper, respectively). 341 342 Empirical detection limits. The linear relationship between log10 -transformed incidence rate and 343 log10 -transformed N cp/g for each plant was used to estimate the incidence rate detection limit 344 assuming an assay detection limit of 750 cp/g. The minimum number of estimated cases 345 detectable in each sewershed was calculated based on the population served by each POTW. 346 Across the eight POTWs, the average estimated incidence rate detection limit was 1.4 cases 347 per 100,000 people (range 0.8 -2.3 cases per 100,000 depending on POTW; Table 3). In the 348 POTW service areas included in this project, these rates correspond to between 1.0 and 26.3 349 cases per sewershed depending on sewershed (Dav and Sac are represented by the minimum 350 and maximum report in the range, respectively). This estimation is corroborated by 351 measurement of N gene at concentrations above 750 cp/g when there were 1.3 recorded cases 352 at Dav although data in this study does not reach the detection limit and future data can lend 353 more insight to these predictions. 354 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. individuals may be infectious before onset of symptoms or asymptomatic such as COVID-19 360 and norovirus, among others, tests may not be sought or may be unavailable. During the 361 COVID-19 pandemic, racial and ethnic minority groups in the United States have been at higher 362 risk of morbidity and mortality in part due to lack of access to testing and care 24 . Wastewater 363 monitoring for SARS-CoV-2 RNA can provide an estimate of COVID-19 incidence rates in 364 communities that are unbiased by these factors. For example, apparent dips in incidence rates 365 associated with testing bias during the holidays were not reflected in wastewater trends. We 366 modified an academic laboratory protocol for measuring SARS-CoV-2 RNA in wastewater solids 367 so that it could be executed quickly (in less than 24 hours) and with numerous samples in 368 parallel. Using the high-throughput and rapid protocol, we provided daily measurements of 369 SARS-CoV-2 RNA and associated quality assurance control metrics to stakeholders on a daily 370 basis via a public website (wbe.stanford.edu). Over the period of more than 4 months 371 represented in this analysis, sample results were always available within 24 h of sample 372 retrieval from the POTW with the exception of holidays. POTW staff rarely missed sample 373 collection. 374

375
The strong association between SARS-CoV-2 RNA in solids and incident case rates within 376 sewersheds, and across sewersheds is striking. The nearly linear relationships between these 377 measures, whether they are examined for individual POTW or data aggregated across all 378 POTW, provide evidence that measurements of SARS-CoV-2 RNA in wastewater solids reflect 379 trends in COVID-19 incidence rates for the sewershed. Importantly, associations are strong for 380 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ; https://doi.org/10. 1101/2021 all POTWs and there is no difference between plants that provided grab samples of settled 381 solids, composite samples of settled solids, or solids settled from composite influent samples. 382 This suggests that even for plants without a primary clarifier, solids represent a viable matrix for 383 wastewater surveillance efforts. While 750 cp/g is the estimated detection limit for the methods 384 used for this data, the detection limit can be lowered even further with some adjustments to 385 these methods.  (Table S2) 397 and the same partitioning coefficients were used in calculation F for each POTW, it is not 398 surprising that the scaling N gene concentrations by F does not appreciably affect associations 399 with incidence rate; this was also reported by Wolfe et al. 7 . 400

401
We suggest that SARS-CoV-2 RNA gene concentrations should be normalized by PMMoV and 402 a trimmed average applied for stakeholder and public consumption of the data. Normalizing by 403 PMMoV serves a number of purposes: (1) it adjusts for variable viral RNA recovery between 404 samples (assuming PMMoV RNA recovery is similar to that of SARS-CoV-2 RNA), (2) it adjusts 405 for differences in fecal strength of the wastestream, and (3) the normalization falls from a mass 406 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 balance model relating the number of shedders to concentrations in wastewater solids 7 . 407 Although normalizing by PMMoV did not improve the associations observed in data from this 408 study, the benefit of this approach is especially realized in our analysis illustrating that results 409 from two distinct laboratories using distinct methods to measure SARS-CoV-2 RNA in solids can 410 be combined when SARS-CoV-2 RNA concentrations are normalized by PMMoV. In this case, 411 the median recovery of BCoV spiked into the solids described by Wolfe et al. 7 was 4% while 412 BCoV recovery in the present study was approximately 10 times higher. As such, normalizing by 413 PMMoV likely served to adjust for variable recovery. Additionally, recent work by Simpson et al. 414 25 suggests that normalizing by PMMoV can serve to correct for degradation of SARS-CoV-2 415 RNA during sample storage. 416 417 A benefit of having daily measurements is the ability to apply a trimmed averaging approach for 418 data visualization. We recommend applying a trimmed average to eliminate the influence of 419 outliers during data visualization by public health professionals and non-experts, including the 420 public. Environmental data exhibit variability that is caused by different factors than those that 421 are most familiar to these audiences, e.g. biases that cause variability in clinical case or 422 syndromic data. SARS-CoV-2 RNA concentrations in wastewater may exhibit high-frequency 423 variability for several reasons. Among the most influential are: 1) changing contributions to the 424 sample from sudden movement of people shedding viral RNA in their stool into or out of the 425 sewershed or intermittent deliveries of septic waste that could vary in content or time 426 represented relative to wastewater flows, 2) variability in fecal shedding rates from person to 427 person 26,27 , and 3) samples are spatially heterogenous and, although our samples are well 428 mixed, inhomogeneities could exist at spatial scales greater than those captured by our 429 sampling. 430

431
Conclusions 432 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. ;https://doi.org/10.1101https://doi.org/10. /2021 We measured three SARS-CoV-2 RNA targets in wastewater solids daily for over 4 months for 433 consumption by POTW and public health stakeholders. The methods used were shown to be 1) 434 sensitive to identify low concentrations of SARS-CoV-2 and COVID-19 incidence rates in the 435 associated sewersheds, 2) scalable to a high-throughput format with results delivered daily 436 within 24 hrs, 3) representative of disease incidence in the sewersheds served, and 4) 437 comparable across laboratories using different methods to analyze solids. POTW staff provided 438 samples and rarely missed a sample. Using a high-throughput method that takes advantage of 439 automation and robotics, we were able to provide sample results within 24 hours of sample 440 receipt and displayed those results on a public website for stakeholders. As the use of 441 wastewater monitoring is poised to scale globally for monitoring not just COVID-19, but also 442 other diseases, it is critical that methods are scalable for use by labs producing reliable results 443 at an industrial scale. Strict QA/QC procedures (including negative and positive extraction and 444 PCR controls for all targets, and recovery controls) coupled to replicate analyses (n=10) for 445 each sample ensured high quality data. Measurements at each POTW are strongly associated 446 with lab-confirmed COVID-19 incidence rates in the sewersheds dated to the time of specimen 447 collection. Further, the strong association persists when data are aggregated across POTW, 448 suggesting that SARS-CoV-2 RNA concentrations in settled solids from different POTW can be 449 directly compared to infer relative incidence rates across POTW sewersheds. Although 450 normalizing data by PMMoV was not necessary for comparing measurements made at different 451 POTW in this study, we show how normalizing by PMMoV can allow for data from wastewater 452 solids measured by different methods and labs to be readily compared to each other. Public 453 health representatives have expressed an interest in comparing these data across sewersheds. 454 Future work will utilize a longer time series of daily wastewater measurements from these 455 POTWs to investigate the appropriate cadence of sampling to capture incident case trends 456 during different phases of the pandemic, and utilize solids settled from samples capturing sub-457 sewershed areas to illustrate use of these methods at different scales. 458 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted July 20, 2021. column is the number of samples analyzed at each POTW. Information on the residence time of 605 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 20, 2021. ; https://doi.org/10. 1101/2021 sewage in the network and solids in the clarifier was obtained through a survey completed by 606 POTW operators and managers. 607 608 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 20, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 20, 2021.   . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted July 20, 2021. ; https://doi.org/10. 1101/2021