Emerging infectious diseases (EIDs), such as COVID-19, represent a persistent global health challenge, characterized by their rapid spread and significant societal impact (1–4). Defined as infections that have newly appeared or increased in incidence within the last two decades, EIDs like SARS-CoV-2 pose a critical burden on health systems worldwide (5–7). These diseases often overwhelm healthcare infrastructure, necessitate substantial investments in treatment and vaccination, and exacerbate pre-existing health disparities (8, 9). In Mexico, the COVID-19 pandemic has highlighted vulnerabilities in healthcare access and response capabilities, particularly in densely populated urban centers (10–16).

The SARS-CoV-2 virus, responsible for COVID-19, spreads primarily via respiratory droplets but has also been detected in fecal matter, suggesting wastewater as a potential surveillance medium. Wastewater-based epidemiology (WBE) has gained traction as a non-invasive tool to monitor viral prevalence within communities (17–21). Unlike traditional epidemiological approaches, WBE provides a cost-effective and comprehensive snapshot of population-level infection dynamics, encompassing symptomatic and asymptomatic cases (22–24). Such methodologies are particularly relevant in low-resource settings where widespread clinical testing may be unfeasible.

Globally, WBE has proven effective in early outbreak detection, guiding public health interventions, and estimating disease prevalence (25, 26). The approach offers distinct advantages for public health and policy. First, it enables real-time monitoring of infection trends at a community scale, reducing the need for costly individual-level testing. Second, WBE can identify asymptomatic carriers who might otherwise go undetected in traditional surveillance systems (27–29). In parallel, WBE has enabled the early identification of emerging variants, often several days or even weeks ahead of clinical sampling, underscoring its potential utility for genomic surveillance (30–33). Finally, it supports proactive health policy decisions by providing data that inform resource allocation, intervention timing, and public communication strategies. These advantages make WBE an invaluable component of integrated public health surveillance systems (34, 35).

In Mexico City, one of the areas most severely impacted by COVID-19, leveraging WBE presents a unique opportunity to strengthen epidemiological surveillance and inform policy decisions (12, 13, 36, 37). This study evaluates the temporal relationship between SARS-CoV-2 RNA levels in wastewater from the Copilco neighborhood and reported infection trends, aiming to validate WBE's role in public health preparedness and response.

A diagrammatic scheme with the methods described here is presented in the form of a workflow in the Figure 1.

This study reinforces the value of wastewater monitoring as a surveillance tool, particularly in urban settings. The significant time lags observed indicate that wastewater monitoring can provide early warnings of infection spikes, enabling proactive public health responses. While this study focused on a single sampling site, the surrounding population ranges from 8,000 to 43,000 inhabitants, depending on the zone (A, B, C) analyzed. This is a strategic location, as it serves as a wastewater collection point for the surrounding neighborhoods. Due to these characteristics, the study assumes high local representativeness, but expanding the sampling network to increase coverage and reliability could improve the representativeness of Mexico City.

It should be noted that a formal power analysis was not performed due to the lack of standardized methods for estimating power in time-lagged WBE studies of SARS-CoV-2 viral RNA concentrations at the time of sampling. This work should be considered an exploratory analysis demonstrating the feasibility of detecting temporal correlations between SARS-CoV-2 N1 gene quantifications in wastewater and reported infections in a specific urban context. Future work would benefit from larger sample sizes and power calculations based on the effect sizes and time lags observed here. Furthermore, the detection of SARS-CoV-2 RNA in wastewater only reflects the presence of genomic material and does not guarantee viral viability (59–61). Therefore, the observed peaks should be interpreted as a population indicator of epidemiological trends useful for monitoring but not as a direct measure of the risk of community transmission.

To address missing data from dates without sample collection or reported cases, data interpolation was performed for both datasets. A higher number of interpolated values were required for wastewater samples, though an expected trend of genome count fluctuations was observed. Additionally, data smoothing was applied to reduce noise and facilitate subsequent analysis, ensuring the identification of generalizable patterns within the model's variable constraints.

To conduct an exploratory graphical analysis of the potential temporal relationship between both datasets, a scatter plot analysis was performed. This involved shifting the genome count data by one-day increments relative to the number of reported cases, within a time-lag window ranging from 14 days before to 5 days after. The analysis was carried out using the complete dataset from both sampling periods combined (TB1) and separately for each sampling period (TB2 and TB3). As it was previously mentioned, a cross-correlation analysis was performed. In all cases, the cross-correlation factor (CCF) is higher for TB2 and TB3, that is, when the sampling seasons are analyzed separately. It is also greater for Zone C, which corresponds to the sum of infection cases reported in Proximity Levels 1, 2, and 3. Additionally, the time lags with the highest CCF values correspond to the scatter plots with the steepest slopes. This suggests that the relationship improves when analyzing time periods with better representativity of sampled days and a higher number of reported cases. Therefore, these two parameters were continuously refined to enhance the proposed model.

For TB2 (April 2021 to September 2021), the highest CCF (0.823 to 0.825) corresponds to a lag of 7 to 9 days before the reported cases in the population. In contrast, for TB3 (November 2021 to February 2022), the highest CCF (0.918 to 0.924) corresponds to a lag of 5 to 7 days before the reported cases. As it was shown in the Results section, these time windows align with the steepest slopes in the previous exploratory graphical analysis, indicating the period before detection through direct epidemiological evaluation in the population.

One of the first studies conducted to evaluate the correlation between the presence of SARS-CoV-2 genomes in wastewater and reported cases in the population analyzed data from treatment plants in six cities and an airport in the Netherlands. It found the presence of viral particles 7 to 9 days in advance using RT-qPCR (56). Another study evaluating 32 treatment plants in Catalonia, Spain, detected viral genomes 7 days in advance. It also assessed different population sizes and models, concluding that understanding population dynamics can lead to a more accurate predictive model (57). In a separate study that collected a total of 1,101 samples from various treatment plants and sewer systems in Paris, France, it was found that SARS-CoV-2 genomes could be detected 3 to 4 days in advance, particularly in populations with limited mobility. The variation in detection also depended significantly on the time of sample collection mainly due to the different dynamics of population behavior (62, 63).

A more recent study conducted in Xàtiva, in the province of Valencia, reported predictive windows of 15 to 17 days using different models and hospitalization data. This was facilitated by a more established local sampling strategy and a well-documented population dynamic (64). Meanwhile, another study in Yamanashi Prefecture, Japan, reported a predictive window ranging from 3 to 9 days (58).

It is important to note that some longitudinal studies indicate that a proportion of those infected shed SARS-CoV-2 RNA in feces days before the onset of symptoms, but shedding can also last several weeks, so the exact magnitude of this presymptomatic phase varies between populations and viral lineages (65–67). Therefore, inference of time lags from WBE should be interpreted with caution due to this uncertainty, and the set-up of a monitoring system should consider these issues (61, 68).

A key aspect of this study was the selection of populations associated with the sampling points. This was done by integrating information on drainage systems with population sizes linked to postal codes, defining an area of influence around the selected collection point. This approach allowed for the delineation of reported infection data corresponding to the sampling site. This is particularly important because infection dynamics may differ in other areas, and including additional populations could increase variability in the model, making it less robust.

All these findings indicate that predictive time windows can vary across different locations depending on the population and site context. The inclusion of intrinsic variables related to the population's dynamics, along with a detailed understanding of wastewater characteristics and additional health system data, could help refine these predictive models. In the case of this study, the determination of the geographic area and, consequently, the study population was appropriate; however, incorporating a parameter to normalize the size population size associated with each sample could help reduce result variability if applied correctly. Therefore, it is recommended to explore measurement strategies such as genes associated with Pepper mild mottle virus (PMMoV) or physicochemical parameters like chemical oxygen demand (COD) or different nitrogenous compounds to include them as part of the analysis process (69–72).

The findings of this study highlight the effectiveness of wastewater-based epidemiology as a viable tool for monitoring SARS-CoV-2 infection trends at the community level.

The optimized RT-qPCR method for quantifying the N1 gene of SARS-CoV-2 reported in this study has a detection limit of 10 copies/μL with a runtime of 30 to 35 min using separately available reagents. Additionally, it can be optimized for even smaller volumes or adapted to different enzyme brands if necessary. This makes it a viable alternative to commercial test kits, which can be more expensive or have limited availability.

The various sample processing methods used for detecting SARS-CoV-2, along with inherent variations in population behavior, contribute to variability in the results. Despite this, the raw data suggest that infection trends in the target population–both increases and decreases–can be tracked using this direct wastewater monitoring method integrating information on drainage systems with local population size. Even with a relatively small number of samples, this approach demonstrates effectiveness compared to traditional epidemiological monitoring methods, however, incorporating a method to normalize population size would be necessary to further improve the model.

When comparing the genome count data for the N1 gene with reported infection data, the application of computational methods for data interpolation on missing dates, along with data smoothing using the central simple moving average technique, revealed a time lag of 5 to 9 days. Additionally, a cross-correlation factor ranging from 0.825 to 0.924 was observed between the genome detection curves and reported infection curves. This lag can serve as an early warning for infections in monitored populations, allowing for the implementation of public health contingency measures if needed.

The method presented here can be replicated in other populations, provided that sampling points in the sewage system are carefully selected. Combined with a well-planned collection schedule, this approach can help validate and improve the proposed model.

Both the sample processing techniques and computational analysis methods can be further refined through continuous feedback, which would enhance the effectiveness of the proposed monitoring system.

In conclusion, we have shown how WBE offers a viable approach for monitoring SARS-CoV-2 and potentially other pathogens. Future research should integrate more variables, such as mobility patterns and climatic factors, to refine predictive models and enhance public health interventions.