Novel methods for global water safety monitoring: comparative analysis of low-cost, field-ready *E. coli* assays

Water quality monitoring can serve as a critical feedback mechanism to support safe water supplies and public health, yet in many low- and middle-income countries microbiological testing is constrained by the need for hygienic laboratory space, specialized equipment, costly consumables, and trained personnel. These constraints often coincide with areas of greatest water safety need, creating a pressing need for simple, scalable E. coli tests usable by non-experts. Simpler, potentially low-cost alternatives to standard membrane filtration assays are available for detection of E. coli and other faecal indicator bacteria, with some supported by systematic comparative testing. Using criteria of total cost per test of ≤US$2 per sample and a lower limit of detection of 1 cfu E. coli in 100 ml by culture, the study selected two novel assays—CompactDry™ (modified membrane filtration with dehydrated media plates) and Aquatest (AT) presence-absence tests at 10 ml and 100 ml—for evaluation against EPA Method 1604 (MF-MI). The hypothesis was that these low-cost assays would achieve high (≥90%) sensitivity and specificity relative to the reference method under both standard and ambient incubation conditions, supporting scalable monitoring toward SDG 6.

The study situates its work within efforts to identify low-cost, scalable microbial water tests suitable for LMICs, noting systematic comparative testing of alternative assays and catalogs of tests for low-resource settings. Prior work demonstrates feasibility of ambient-temperature incubation for E. coli detection and evaluates AT against standard methods in environmental waters, reporting high sensitivity but some false positives. The paper references challenges and costs of monitoring at scale, and previous evaluations of related quantal and H2S-based tests, highlighting gaps the present evaluation addresses (field-ready, low-cost, sensitive and specific assays compatible with SDG monitoring).

Study setting and sampling: Drinking water samples were collected from 14 tap locations in Peenya, Bangalore, India, across 37 non-consecutive days in 2017–2018. Peenya has ~800,000 residents and ~1300 people per km². At each point, three water samples totaling ~2000 ml were collected into sterile Whirl-Pak bags with sodium thiosulfate, kept on ice, and analyzed within 5 h. In total, 315 bulk tap water samples were each assayed in triplicate by all methods (n=945 assays per method). Reference method (MF-MI): Membrane filtration (47-mm diameter, 0.45 µm cellulose ester filters) followed by incubation on MI agar for 24 h at 35 °C per EPA Method 1604. MI agar uses chromogenic/fluorogenic substrates for total coliforms and E. coli; E. coli colonies fluoresce under 366 nm UV. Results were reported as cfu per 100 ml; upper limit of quantification was 200 cfu per plate (TNTC above this). Aquatest (AT) presence-absence tests: Open-source AT broth with resorufin methyl ester chromogen in single-use volumetric cylinders for 10 ml and 100 ml tests. Incubation at 37 °C for 24 h or at ambient temperature (mean 27 °C; range 25–30 °C) for 48 h. Color change from yellow-beige to pink-red indicates E. coli presence. Combining 10 ml and 100 ml results yields categorical risk: <1 E. coli/100 ml (both negative); 1–10 E. coli/100 ml (100 ml positive, 10 ml negative); ≥10 E. coli/100 ml (10 ml positive). CompactDry™ assay: Modified method involving membrane filtration of 99 ml (for 100 ml assay) onto 47-mm filter, placed on pre-sterilized CompactDry™ plate with dehydrated medium, rehydrated with 1 ml sample water. For 10 ml tests, 9 ml filtered and plate rehydrated with 1 ml. Incubation for 24 h at 35 ± 2 °C and 48 h at ambient temperature (25–30 °C). E. coli and total coliform counts reported as cfu per 100 ml; TNTC >200 cfu assigned 200 for means. Requires standard filtration equipment but avoids media preparation. Quality control: Negative controls with sterile dilution water were run daily for each method. Statistical analysis: All methods performed in triplicate per sample. Primary analysis assessed comparability across individual replicates. Secondary comparisons used MF-MI triplicate means as the reference per sample. Sensitivity, specificity, and PPV for detection of E. coli (≥1 per 100 ml) were calculated with exact Clopper-Pearson CIs (PPV via logit CIs). Descriptive statistics characterized distributions (raw and log10-transformed), with non-detects set to 1 for log10 analyses. Agreement and correlation assessed via scatter plots, Spearman’s correlation, and Bland-Altman plots (mean differences via linear regression accounting for replicates). ROC curves assessed diagnostic performance as MF-MI thresholds varied from ≥1 to the upper limit; area under the curve (AUC) and CIs accounted for clustering by sample.

- Sample characteristics (MF-MI reference): Arithmetic mean total coliforms 70 cfu/100 ml (SD 82); mean E. coli 57 cfu/100 ml (SD 75); range <1 to 200 cfu/100 ml (upper limit), so means are underestimates. By MF-MI, 35% of samples were <1 E. coli/100 ml (safe) and 26% were 101+ E. coli/100 ml (high risk). - ROC performance: AUC for AT exceeded 0.97 under both incubation regimes; AUC for CompactDry™ ≈0.99 (near-ideal) versus MF-MI. Reported examples include AUC: AT incubated 0.971 (95% CI 0.957–0.986); CompactDry incubated 0.973 (95% CI 0.959–0.987); CompactDry ambient 0.992 (95% CI 0.987–0.998). - CompactDry™ accuracy (single plate counts, n=945): Sensitivity 99.5% (95% CI 98.6–99.9) for both standard and ambient incubation; specificity 97.3% (95% CI 94.9–98.8) for both; PPV 98.6% (95% CI 97.3–99.2). Agreement with MF-MI risk categories: <1 cfu: 97%; 1–10: 40–45%; 11–100: 83–84%; 101+: 95%. Underestimation 2.2–2.3% of samples; overestimation 11–12%. - AT presence-absence accuracy (n=945): • 100 ml test: Sensitivity 97.1% (95% CI 95.4–98.2) at 37 °C/24 h and 96.7% (95% CI 95.0–98.0) at ambient/48 h; specificity 95.8% (95% CI 93.1–97.7) and 96.7% (95% CI 94.2–98.3); PPV 97.7% (95% CI 96.2–98.6) and 98.2% (95% CI 96.8–99.0). Underestimation 1.9–2.1%; overestimation 1.2–1.5%. • 10 ml test: Sensitivity ~51% at both temperatures (51.4% and 51.6%); specificity ~100% (99.8% and 100%); high underestimation of risk (~25%). Nearly half (49%) of samples were negative when MF-MI counts were ≥11 cfu/100 ml. - Quantitative agreement (CompactDry vs MF-MI): Strong correlations across conditions. Spearman’s rho: 0.982 (incubated vs reference), 0.986 (ambient vs reference), 0.991 (incubated vs ambient). Bland-Altman mean differences (log10 CFU/100 ml): CompactDry incubated vs MF-MI 0.003 (95% CI −0.067 to 0.118); CompactDry ambient vs MF-MI 0.100 (0.078–0.122); CompactDry incubated vs ambient 0.001 (−0.01 to 0.018). - Incubation conditions: Comparable sensitivity/specificity at ambient temperatures (25–30 °C, mean 27 °C) for both methods relative to standard incubation.

The study’s hypothesis that low-cost assays could achieve high sensitivity and specificity relative to MF-MI was supported. Both CompactDry™ and AT 100 ml presence-absence tests provided highly sensitive and specific detection of E. coli in 100 ml drinking water samples, under both standard incubation and ambient conditions (25–30 °C). CompactDry™ also showed strong quantitative agreement with MF-MI across the full range of risk categories, indicating suitability where quantitative counts are needed. The findings indicate that ambient-temperature incubation within the tested range can yield equivalent results to standard incubation, potentially enabling simpler field workflows without specialized incubators. The poor performance of the 10 ml AT test underscores the need for adequate sample volumes in presence-absence testing; smaller volumes may fail to reliably detect moderate contamination (≥11 cfu/100 ml). These results are pertinent to SDG 6 monitoring, suggesting feasible, lower-cost options for E. coli surveillance in resource-limited settings. Comparisons with prior work show consistency in AT sensitivity and highlight that ambient incubation performance may depend on maintaining temperatures ≥25 °C.

This study identifies two low-cost, field-ready E. coli assays suitable for scalable water quality monitoring. CompactDry™ (with membrane filtration) shows near-ideal diagnostic performance and strong quantitative agreement with MF-MI, making it appropriate where quantitative counts are required. The AT 100 ml presence-absence test provides highly sensitive and specific detection suitable for rapid, field-ready surveillance when quantal data suffice. Ambient incubation at 25–30 °C for 48 h produced results comparable to standard 37 °C/24 h incubation for both methods, enabling use in basic settings without incubators. Tests using <100 ml (e.g., 10 ml) exhibited inadequate sensitivity and are not reliable for assessing drinking water safety where the standard is absence of E. coli in 100 ml. Future research directions are not explicitly stated.

- Upper limit of quantification was 200 cfu per plate (TNTC above this), leading to underestimation of true means at higher contamination levels. - Ambient incubation evaluations were conducted at 25–30 °C (mean 27 °C); performance at lower ambient temperatures may differ, as suggested by other studies, implying a minimum recommended threshold of ~25 °C for ambient AT incubation. - Testing focused on drinking water taps in one urban area (Bangalore) and on 100 ml volumes for primary comparisons; results may not generalize to other water types or volumes without further evaluation.