During October 29, 2023–May 31, 2025, a total of 801 wastewater samples were collected across 16 WWTPs serving an estimated population of ≈2 million people. Of those, 47 (5.9%) were classified as wastewater-positive (17 above the LOD and 30 trace detections) and 754 (94.1%) were wastewater-negative. During the same period, 45 confirmed mpox cases were reported that fell within sewersheds representing the WWTPs in this study.

Using a +17-day temporal alignment window, we classified 15 wastewater observations as TPs, 32 as FPs, 627 as TNs, and 127 as FNs (Table 2). The overall sensitivity of wastewater detection identifying the presence of >1 temporally aligned reported mpox case within a sewershed was 10.6% (95% CI 6.5%–16.7%). Specificity was 95.1% (95% CI 93.2%–96.5%). The PPV was 31.9% (95% CI 20.4%–46.2%), and the NPV was 83.2% (95% CI 80.3%–85.7%).

At the WWTP sewershed level, sensitivity was highest at Charlotte WWTP 2 (25.0%) and ranged from 0% to 25.0% across sites with clinical activity (Table 3). Several sewersheds had no temporally aligned clinical cases during the surveillance period, resulting in undefined sensitivity estimates. Specificity was consistently high across sites, ranging from 71.4% to 100%. PPV varied widely by facility, ranging from <1% to 100%, reflecting differences in case occurrence and sampling frequency. NPV exceeded 69% at all sites with clinical activity and was 100% at several facilities without aligned cases.

When restricting wastewater-positive observations to samples exceeding the LOD (n = 17), sensitivity decreased from 10.6% (95% CI 6.5%–16.7%) to 1.4% (95% CI 0.4%–5.0%), and specificity increased from 95.1% (95% CI 93.2%–96.5%) to 97.7% (95% CI 96.3%–98.6%) (Table 2). The number of FP observations decreased from 32 to 15, whereas FN observations increased from 127 to 140. PPV decreased from 31.9% (95% CI 20.4%–46.2%) to 11.8% (95% CI 3.3%–34.3%), and NPV was similar (83.2% vs. 82.1%). Site-level performance patterns were unchanged.

This evaluation assessed the performance of wastewater surveillance for detecting MPXV across 16 WWTPs in North Carolina during October 2023–May 2025. Using a +17-day alignment window, wastewater detection demonstrated low sensitivity for identifying temporally aligned clinical cases (10.6% overall when including trace detections and 1.4% when restricted to detections above the LOD) but high specificity (>95%) and moderate NPV of ≈83%. We selected a +17-day window to account for the expected duration of MPXV shedding and potential delays between symptom onset, clinical diagnosis, and case reporting, while also accommodating once-weekly wastewater sampling. Because wastewater represents a composite, time-integrated signal and samples are collected weekly, this window reduces misclassification that could arise from temporal offsets between individual infections and wastewater detection.

Those findings compare with national estimates reported in another study (16) that observed a daily sensitivity of 13.8% for detecting >1 person shedding MPXV and a weekly sensitivity of 31.7%. The lower sensitivity observed in North Carolina likely reflects differences in analytic approach, case density, and assumptions about viral shedding. Adams et al. (16) assumed complete case ascertainment and uniform 25-day viral shedding after symptom onset of mpox, whereas our analysis relied on confirmed case onset dates and a fixed +17-day temporal alignment window. This approach likely provides a more conservative estimate of sensitivity because wastewater detections occurring outside the defined window were classified as discordant even if they reflected prolonged or presymptomatic shedding. In addition, only 45 cases were identified across a population of >2 million persons, limiting the probability that wastewater concentrations would exceed detection thresholds.

Despite low sensitivity for detecting single cases, specificity was high (95%–98%), indicating that wastewater-negative observations were generally concordant with the absence of reported cases. The NPV exceeded 80% overall and was higher at several individual sites, suggesting that the absence of a wastewater detection was associated with a low likelihood of concurrent reported cases. However, the PPV was modest (31.9% in the primary analysis and 11.8% in the LOD-only analysis), reflecting that some wastewater detections occurred without temporally aligned reported cases. Such detections could represent travel-associated cases, delayed reporting, prolonged shedding, asymptomatic or subclinical infections, or infections among persons residing outside the mapped catchment area. As a consequence, wastewater detections occurring without temporally aligned reported cases should not necessarily be interpreted as analytical FPs and might indicate true but unrecognized infections, suggesting that PPV might be underestimated. Therefore, if even a subset of those observations represented true infections, the PPV of wastewater surveillance would be higher than estimated in this analysis.

Figure

Figure. Public Health Action Decision Tree for monkeypox virus surveillance in wastewater, North Carolina, USA, October 2023–May 2025. Period of 28 days is based on typical lesion healing duration (16...

Of note, sensitivity increased substantially when trace detections were included, indicating that low-level signals contributed meaningfully to temporal concordance with reported cases. That finding underscores the tradeoff between sensitivity and specificity when defining wastewater-positive thresholds. Restricting classification to concentrations above the assay LOD generally increased specificity but markedly reduced sensitivity, potentially limiting early warning capacity in low-incidence settings. In North Carolina, trace detections are interpreted within a structured public health decision framework rather than as definitive evidence of transmission and might prompt notification of local health departments and wastewater partners, review of recent case activity, enhanced syndromic surveillance monitoring, and coordination regarding case investigation or specimen retesting. Those actions are consistent with our public health action decision tree for confirmed detections (Figure). Confirmed detections might also involve escalation actions such as broader public notification, targeted outreach, and intensified sampling.

Several factors likely contributed to the low sensitivity observed. First, mpox incidence during the study period was low and sporadic, reducing the probability that viral shedding might be detectable within large, pooled wastewater catchments. Second, sampling occurred once per week, which might have missed transient shedding signals. Third, wastewater represents an aggregate population-level signal, and the dilution effect within large catchments (some serving >500,000 persons) might limit detection of individual infections. Fourth, alignment relied on residential address assignment; persons infected but not residing within the sewershed would reduce measured concordance.

The first limitation of this study is that reported clinical cases are not a perfect standard for infection. Our intent was to quantify the extent to which wastewater detections were temporally aligned with reported clinical cases within the same sewershed, rather than to provide an estimate of the true diagnostic accuracy of wastewater surveillance for detecting infection. Second, clinical case reporting was assumed to be complete within each sewershed; however, underascertainment of cases could bias estimates of sensitivity and PPV downward while inflating NPV. Several factors might have contributed to incomplete case detection, including mild or atypical symptoms, barriers to healthcare access, stigma, variation in clinical testing practices, and infections occurring among visitors or commuters whose residence falls outside the wastewater catchment. As a result, wastewater detections might occur in the absence of a temporally aligned reported clinical case.

An additional limitation of this study is that viral shedding dynamics were not directly measured, and the +17-day alignment window might not fully capture the duration or variability of shedding. Wastewater sampling was also conducted only once per week, which might have reduced the likelihood of capturing transient or low-level shedding events and could have contributed to lower observed sensitivity. Finally, although laboratory methods were consistent within this program, sampling frequency and analytic approaches might differ from other surveillance systems and studies, limiting direct comparability with published estimates, including those reported by Adams et al. (16).

Although sensitivity for detecting single cases was limited, high specificity and moderate NPV indicate that absence of detection might provide reassurance that widespread community transmission is unlikely. In low-prevalence contexts, isolated wastewater detections should be interpreted cautiously and in conjunction with epidemiologic investigation (9). Conversely, sustained nondetection in monitored sewersheds might help inform risk communication and public health response decisions. Continued evaluation across varying transmission intensities and analytic thresholds will help clarify the optimal role of mpox wastewater surveillance in outbreak preparedness and response. Finally, because clinical surveillance does not capture all mpox infections, some wastewater detections classified as FPs might instead represent true but unrecognized infections. Consequently, the sensitivity and positive predictive value reported here should be considered conservative estimates of concordance with reported clinical cases rather than measures of true diagnostic accuracy.