Parallel testing between the two-phase separation and CaFÉ methods was conducted at CDC-Atlanta, using samples collected from Haïti between December 2017—December 2019 (n = 144), twelve sampling sites in four coastal cities in Haïti—Port au Prince, Gonaïves, Saint Marc, and Cap-Haïtien were selected (Alleman et al., 2021a) (Table 1; Figure 1A), and samples from Guatemala from November 2018—August 2019 (n = 53). Seven sampling sites were selected in two major cities in Guatemala—San Juan Sacatepéquez, a municipality located in the northwest of Guatemala City. Villa Nueva located south of Guatemala City (Table 1; Figure 1B). These two cities were selected for evaluation, according to the GPEI environmental surveillance guidelines (Organization, 2003), based on 1) population, 2) and road accessibility during the rainy and dry seasons.

Geographical coordinates for each site were measured by a Montana 600, a handheld GPS device (Garmin International, Olathe, KS, Unites States), and watershed populations were estimated using the Worldpop spatial demographic dataset (Department of Geography and Environment, University of Southampton, United Kingdom) (Novel-t, 2021).

Six sites were selected in Thailand from previously established sites for poliovirus surveillance (Table 1; Figure 1C) with sample collection conducted between February 2019- July 2019 (n = 58) and processing at the National Institute of Health Laboratory (NIH). Two sites were selected in the country`s capital Bangkok: the Din Daeng site (in the inner zone of Bangkok) and the Nong Khaem site (in the western zone of Bangkok). Two sites were selected from the Udon Thani province: Nongsim and Huay Mak Khaeng and two sites were selected in Mae Sot district, in Tak province at the border with Myanmar. This district is known for the presence of a large hospital and is a transportation hub between the two countries.

Three collection sites were included for the pilot study at the Research Institute for Tropical Medicine Laboratory (RITM) in the Philippines, two in Metro Manila: one in Quezon City and the other one located in the city of Manila. The third collection site is in Baguio City. Samples were collected from October 2018—December 2020 (n = 54) (Table 1; Figure 1D).

Three collection sites were selected in Papua New Guinea: in Port Moresby, Gerehu Sewage Lagoon (GSL), Waigani Sewage Lagoon (WSL) and Joyce Bay Treatment Plant (JTP). The samples were shipped to RITM for processing and included in pilot testing collected from October 2018—December 2019 (n = 29) (Table 1; Figure 1E).

Two 1-L wastewater samples were collected approximately once every 4 weeks at each sampling site in Haïti, Guatemala, Thailand, Philippines, and Papua New Guinea, using the GPLN/WHO recommended sampling method, the grab method with a swing sampler (NASCO, Fort Atkinson, WI) with 1-L Nalgene bottles (Organization, 2015c). In the grab method, the sample was collected at one point in time, also it is a more quantitative method that allows estimation of the system`s detection sensitivity (Organization, 2003). Time, date of collection, and sample temperature as well as weather conditions on collection day and previous day were recorded. All samples were stored at 2–8°C immediately after collection and transported to the laboratory or stored at ≤ −20°C until shipment. All specimens were shipped frozen on dry ice and stored at ≤ −20°C until processing.

Upon arrival at the laboratory, sample temperature was recorded using a temperature gun (Etekcity, Anaheim, CA, United States), and duplicate 1-L environmental water samples for each collection site and collection month were thawed at room temperature (25°C) for 24 h before processing, combined, and mixed for 15 min. Water quality was analyzed using 30 ml aliquot of each sample, by measuring the pH, total dissolved solids and salinity. Two aliquots of 500 ml were measured for processing with two-phase separation and CaFÉ methods, remaining sample volume was stored at −20°C for additional testing if needed.

The samples were concentrated using the two-phase separation method as described previously (Organization, 2015a). Briefly, the sample was centrifuged for 20 min at 6,500 x g at 4°C. The supernatant is added to a beaker and pH was adjusted to 7.0–7.5. The resulting pellet, if any, is stored at 4°C for later addition to the concentrate. Two polymers were added to the clarified sample: polyethylene glycol (6,000) (29% w/v; Sigma-Aldrich, St. Louis, MO) and dextran T40 (22% w/v; Pharmacosmos, Holbaek, Denmark), as well as 5M sodium chloride (Sigma-Aldrich), and then mixed for 1 hour at 10,000 rpm at 4°C. the obtained homogenous mixture is added to a glass separatory funnel and left to stand overnight at 4°C, to allow the polymers separation and form two distinct phases. After incubation, the lower phase and interphase were collected into a 50 ml centrifuge tube (approximately 10–15 ml). The stored pellet from the initial centrifugation was suspended in the concentrate, which is then treated with chloroform (20% v/v; Sigma-Aldrich), agitated for 20 min at 10,000 rpm at room temperature (25°C), then centrifuged for 20 min at 1,500 x g at 4°C. Antibiotics (100 IU/ml penicillin and 100 μg/ml streptomycin and 50 μg/ml gentamycin) were added to the isolated supernatant. The concentrate was inoculated into cell culture for poliovirus isolation the same day, as described below.

The CaFÉ method is based on the very known principle of charge-based methods, where electrostatic forces are instrumental in virus-filter interactions. A volume of 500 ml was processed using the CAFÉ procedure, which utilizes a 1-L, stainless-steel coffee press (Vonshef, London, United Kingdom) using 11 µm cellulose filter papers (85 mm, grade 1, Cytiva Life Sciences, Hillerød, Iceland) that is placed in between the spiral plate and the mesh screen (Figure 2). Wastewater was added to the carafe and then pressed to separate the heavy sediment from the liquid. As shown in Figure 3, viral particles were extracted with beef extract (3% w/v, pH 7.2 ± 0.2; Criterion, Hardy Diagnostics, Santa Maria, CA, United States) and chloroform-dithizone (0.001% w/v; Sigma-Aldrich) from the sediment and the 11 µm filter that were placed in a 50 ml conical tube, then agitated at 10,000 rpm for 10 min at room temperature (25°C) using Heidolph shaker (Schwabach, Germany) and centrifuged for 10 min at 1,500 x g at 4°C. The resulting supernatant was added to the main pressed sample. Magnesium chloride hexahydrate (2.5% w/v; EMD Millipore Corp, Burlington, MA, United States) was added to a final concentration of 0.05 M, and the pH was adjusted to 3.5 to facilitate and optimize virus adsorption to filter surfaces. The pressed sample-supernatant mixture was then passed through a series of two additional negatively charged mixed cellulose filters with a diameter of 47 mm (Advantec, Toyo Roshi Kaisha, Ltd., Uchisaiwaicho, Chiyoda City, Japan), using a vacuum pump (Cole-Parmer, IL, United States). If the sample is turbid with heavy sediment, 25 ml of the sample liquid is added at a time to the filter funnel. The first stage filtration uses a 5 µm filter, which captures virus particles aggregating to large clumps of sediment. During the second stage filtration, filtrate is then passed through a 0.45 µm filter, which captures virus particles aggregating to finer sediment. Both filters were subsequently cut into four pieces and placed in a 50 ml conical tube containing beef extract (3% w/v, pH 7.2 ± 0.2), which will slightly change the virus`s natural charge (or that of the filter) to facilitate virus elution from the filter, and agitated with heidolph shaker for 20 min, at 10,000 rpm at room temperature (25°C). The resulting eluate (15–18 ml) was treated with chloroform (20% v/v), agitated for 20 min at 10,000 rpm at room temperature (25°C), and then centrifuged at 1,500 x g for 20 min at 4°C. Antibiotics (100 IU/ml penicillin, 100 µg/ml streptomycin, and 50 µg/ml gentamycin) were added to the concentrate, then inoculated into cells for poliovirus isolation the same day.

From April 2018 through November 2019, eight Sabin-Like type 3 polioviruses (SL 3 PVs) were isolated from samples collected from seven sites (Bois de chêne (BDC), Bois de Neuf (BNF), Boulevard de l`Avenir (BRA), Ruelle Caporis (CRC), Rue Pétion (PET), Route Rails Diquini (RRD) and Avenue Maurepas (AMA)). Sabin-Like type 1 polioviruses (SL 1 PVs) were only isolated from RRD on April 2018 and from BNF in December 2019, using the CaFÉ method. The same samples were NPEV or not detected with the two-phase separation method. The two-phase separation method detected four SL 3 PVs from four sites (BDC, RRD, AMA and PET), that were not detected using the CaFÉ method (Alleman et al., 2021a) (Supplementary Figure S1).

In January 2019, VDPV1(Accession (ACC) no. MZ313559) was isolated from a sample collected from Aldea Cruz Blanca site (ACB) using the CaFÉ method, and VDPV3 (ACC.no. MZ313558) from a sample collected from Rio Platanitos site (PLA) in March 2019 using the two-phase separation method. Sequencing of the viral protein 1 (VP1) regions of the capsid protein showed 11 nucleotide differences compared to their respective Sabin reference for each one; therefore, both viruses met the definition of VDPV (Burns et al., 2014) (Supplementary Figure S2).

Two-phase separation method detected seven samples with a mixture of Sabin-Like type1 polioviruses (SL1 PVs) and SL3 PVs from ACB, Bodega Municipal (CMB) and Cuidad Quetzal (CQU) sites between November 2018—April 2019. Six of these parallel tested samples from the same collection sites were SL3 PVs with the CaFÉ method and one was NPEV (ACB site). Three samples with a mixture of SL1 PVs and SL3 PVs were isolated with the CaFÉ method from CMB and CQU sites. Two-phase separation method detected four SL3 PVs (CBM, CQU and PLA), that were NPEV with the CaFÉ method.

From November 2018—August 2019, the CaFÉ method detected four SL3 PVs that were NPEV with the two-phase separation method (ACB, Colinas de Villa Nueva (CVP) and PLA) (Supplementary Figure S2).

For the six sites that were chosen by NIH laboratory as part of their in-country poliovirus environmental surveillance, the CaFÉ method detected more SL PVs and NPEVs (Table 2 and 3) than the two-phase separation method. SL 1 PV was detected in five samples using the CaFÉ method from four collection sites (Din Deang, Nong Kheam, Mae Sot Hospital and Huay Mak Kaeng), whereas the two-phase separation method identified either NPEV or not detected for those same samples.

SL 3 PVs isolated from Nong Kheam and Nongsim sites and the mixture of both SL 1 PVs and SL 3 PVs in Nong Kheam were detected by the CaFÉ method but not by two-phase separation method (Supplementary Figure S3).

SL 2 PVs were isolated from samples collected in April and May 2020 at the Manila site (Tondo Sewage Pumping Plant (TSP)) and at Quezon City sampling site (East Avenue Sewage Treatment Plant (ESP)), by both concentration methods. In Manila (TSP site), cVDPV1 were isolated from samples collected in July and September 2019 with both methods. A single cVDPV2 positive sample was detected from the same collection site with the CaFÉ method in January 2020. In Papua New Guinea, mixtures of cVDPV1, SL 1 PVs and SL 3 PVs were detected with both methods in all three collection sites in Port Moresby (Gerehu Sewage Lagoon (GSL), Waigani Sewage Lagoon (WSL), Joyce Bay Treatment Plant (JTP)) in October 2018. cVDPV1 was isolated from GSL site in November 2018 using the CaFÉ method, whereas the result for the same sample was NPEV with the two-phase separation method. Also, SL 1 PVs and SL 3 PVs were detected in all collection sites in the Philippines and Papua New Guinea with a similar rate between CaFÉ and the two-phase separation method (Supplementary Figure S4).