Niger is a landlocked country in sub-Saharan Africa. The country’s area is 1,267,000 km². According to data from the National Institute of Statistics, the population was estimated to be 19,865,068 people in 2016 (https://stat-niger.org/wp-content/uploads/2020/06/TBS_2016.pdf) and 25,369,414 people in 2023 (https://stat-niger.org/wp-content/uploads/tbs/tbs_2024.pdf). The country is located in West Africa, between the parallels of 11˚37' and 23˚33' north latitude, and the meridians of 16˚ east and 0˚10' west longitude. Niger has a Sahelo-Saharan climate, which is characterised by high temperatures and low annual rainfall. The climate varies from desert in the north of the country, to semi-arid tropical landscape in the centre, and humid tropical climate in the far south. This retrospective study covered the period from 2016 to 2024 within the framework of environmental surveillance of poliomyelitis in Niger. Environmental surveillance of poliomyelitis was implemented in November 2015, initially with 4 operational sites in Niamey, the country’s capital and was extended in 2016 with the addition of 4 new sites in Maradi and Diffa. Zinder region was included in the ES system with the opening of 2 sites in 2019 year, then Dosso region with the opening of one site in 2023 year. This makes a total of 11 open sites distributed across 5 regions of the country (Figure 1).

In practice one liter (L) of untreated wastewater was collected by grab sampling method from each site twice a month. For each collection, the samples were transported to the CERMES Environmental surveillance Laboratory at 4˚C for laboratory analysis. Approximatively 500 mL of collected samples were processed using the PEG-6000/Dextran concentration method, in accordance with the Global Polio Eradication Initiative protocol [7]. The concentrates were then sent to the World Health Organisation (WHO) Polio Regional Reference Laboratory of Ghana, where tests were carried out to viral isolation, detection and identification of PV and NPEV. Laboratory results were sent to the country every week by the Polio Regional Reference Laboratory.

Laboratory results were stored in the databases of the Nigerien Ministry of Public Health and WHO. To create the database for this study, these two repositories were used to ensure the similarity of the data. Were included in the study database, all wastewater samples correctly processed and identified for which the collection date and laboratory results were available. All duplicates, inconsistent laboratory results between the two repositories, and samples whose processing procedures and identifiers did not comply with WHO standards were excluded from the study database.

Figure 1. Localization of the 11 wastewater collection sites in Niger. The collection sites were Musée National du Niger (MNN), Stade Seyni Kountché (SSK), Ancienne ambassade de Belgique (AAB), Teminus Corniche Fleuve (TCF), Mangué Koira (MGK), Gao Barki (GBK), Soura Bildi (SRB), Jaguindi (JAG), Washington (WAS), Diffa Koura 1 (DK1) and Diffa Koura 3 (DK3).

The meteorological data were obtained from the National Meteorology Directorate. Daily data were obtained from meteorological stations covering the locations of wastewater collection sites throughout the study period. These data refer to the maximum and minimum temperature and relative humidity in degrees Celsius (˚C) and percentage (%) respectively, on the day of wastewater collection; Wind speed in meter per second (m/s) and rainfall in millimeter (mm) on the day of wastewater collection. Wastewater collections were carried out early in the morning at specific times, and it could rain or be windy after sample collection on the same day. This would have no impact on the detection of enteroviruses in untreated wastewater. In order to take this bias into account, average rainfall and average wind speed were calculated based on data from the day of wastewater collection (J0) and the day before (J-1). In Niger, the specific months associated with the rainy season are June, July, August, and September. The months of October to February are associated with the dry and cold season. The months of March to May are associated with the dry and hot season. The seasons were defined according to the associations with the months mentioned above.

The data concerning measured variables were collected over 9 years (2016-2024). Data were aggregated to generate a single data file. The coding of variables and descriptive analysis were carried out using this table. Descriptive, bivariate and multivariate logistic regression analyses were performed using R for Windows version 4.4.3 software, while the graphs were generated using Microsoft Excel 2021. The trend test for enterovirus positivity rate was determined using Epi Info^TM 7 (chi-squared for trend). A p-value ≤ 0.05 was considered statistically significant.

A total of 1989 wastewater samples collected between 2016 and 2024 as part of the Polio surveillance program were included in this study. The descriptive analysis allowed to calculate absolute and relative frequencies of the measured variables. The median, mean, minimum, and maximum were also calculated for each meteorological parameter. Results were summarized on Table 1.

Table 1. Descriptive statistics results.

N = Number of samples tested; n = Percentage of samples tested; NA = Not applicable.

Thus, 1198 wastewater samples tested negative for the presence of enteroviruses, while the number of samples that tested positive was 791. A bivariate analysis (Pearson’s Chi-squared test) was conducted and a significant association was found between the detection of enteroviruses in wastewater and the region or year covered by this study (p-values < 0.001). The findings also demonstrated a significant association between the detection of enteroviruses in wastewater and season or average humidity with p-values < 0.001 for each variable. This was not the same results for average temperature and wind speed variables with p-values of 0.2 and 0.6 respectively. The data on rainfall (daily and the day before) were paired, and the differences between the pairs followed an abnormal distribution. Consequently, Wilcoxon rank sum test (significant if p-value < 0.05) was performed for the numerical variable (“Rainfall average”) to compare the medians between enterovirus negative and positive results. Results showed that there was a significant difference between these median values (p-value < 0.001). Results were summarized on Table 2.

To better illustrate the relationships between each measured variables (Average relative humidity, rainfall, temperature and wind speed) and the detection of enteroviruses in wastewaters, graphs were generated (Figure 2).

The box plot of positive samples was higher for average humidity, suggesting a clear tendency toward higher values compared to negative cases. The median temperatures and wind speed for enterovirus-negative and enterovirus-positive results seemed similar, suggesting that these two parameters did not differ significantly between the two groups.

Also, a histogram was generated to illustrate the positivity rate according to the season. High positivity rate was observed during rainy season (Figure 3).

Then, a trend test was carried out to determine the rate of enterovirus positivity over months during the study’s period. A significant trend test (p-value = 0.005) was obtained after statistical analysis and results showed that the peaks of positivity rate were observed between July and August (Figure 4).

Figure 2. Relationships between measured variables and the detection of enteroviruses in untreated wastewaters. Box plots comparing the (a) average temperature; (b) average humidity; (c) average windspeed for negative and positive results; (d) dot plots comparing the average rainfall for negative and positive results.

Figure 3. Enterovirus positivity rate according to the season.

Figure 4. Evolution of positivity rate over months.

Table 2. Bivariate analysis results.

*p-value: Pearson’s Chi-squared test; Wilcoxon rank sum test for Average rainfall, N = Number of samples tested; n = Percentage of samples tested; IQR = Interquartile range.

A multivariate logistic regression analysis was then conducted to assess the independent influence meteorological parameters on enterovirus detection in untreated wastewater. The Odd Ratio (OR), confidence interval (CI) and p-value were determined. For categorial variables, Year 2016, rainy season and Niamey region were fixed as reference. Multivariate analysis showed that, Average Humidity (OR = 1.02, [1.01; 1.03], p < 0.0001), Maradi region (OR = 2.91, [2.23; 3.81], p < 0.0001) and years 2024 (OR = 4.17, [2.69; 6.53], p < 0.0001), 2023 (OR = 1.61, [1.04; 2.50], p = 0.034), 2022 (OR = 1.62, [1.04; 2.52], p = 0.034) and 2017 (OR = 1.71, [1.08; 2.70], p = 0.022), were associated with higher rates of Enterovirus detection in wastewater. The years 2018 (OR = 0.61, [0.38; 0.98], p = 0.043), 2021 (OR = 0.6, [0.37; 0.97], p = 0.04), Diffa region (OR = 0.73, [0.55, 0.95], p = 0.022) were associated with lower rates of Enterovirus detection in wastewater. The other variables were not associated with the rate of Enterovirus detection of enterovirus in untreated wastewater (p > 0.05). Results were summarized in Table 3.

Table 3. Multivariate logistic regression results.

*OR = Odds Ratio; **CI = Confidence Interval.

The performance indicators of the model used were Accuracy = 70% and the Area Under the Curve (AUC) = 72.9% (Figure 5).

Figure 5. Performance of the multivariable logistic regression mode.