Skip to main content
NCBI home page
Elsevier - PMC COVID-19 Collection logoLink to Elsevier - PMC COVID-19 Collection
. 2022 Oct 19;4:100032. doi: 10.1016/j.heha.2022.100032

Quantifying diurnal changes in NO2 due to COVID-19 stay-at-home orders in New York City

Jenni A Shearston 1,⁎, Ilan Cerna-Turoff 1, Markus Hilpert 1, Marianthi-Anna Kioumourtzoglou 1
PMCID: PMC9580220  NIHMSID: NIHMS1850487  PMID: 36926117

Abstract

Introduction

Policy responses to the COVID-19 pandemic, such as the NY on Pause stay-at-home order (March 22 – June 8, 2020), substantially reduced traffic and traffic-related air pollution (TRAP) in New York City (NYC). We evaluated the magnitude of TRAP decreases and examined the role of modifying factors such as weekend/weekday, road proximity, location, and time-of-day.

Methods

Hourly nitrogen dioxide (NO2) concentrations from January 1, 2018 through June 8, 2020 were obtained from the Environmental Protection Agency's Air Quality System for all six hourly monitors in the NYC area. We used an interrupted time series design to determine the impact of NY on Pause on NO2 concentrations, using a mixed effects model with random intercepts for monitor location, adjusted for meteorology and long-term trends. We evaluated effect modification through stratification.

Results

NO2 concentrations decreased during NY on Pause by 19% (-3.2 ppb, 95% confidence interval [CI]: -3.5, -3.0), on average, compared to pre-Pause time trends. We found no evidence for modification by weekend/weekday, but greater decreases in NO2 at non-roadside monitors and weak evidence for modification by location. For time-of-day, we found the largest decreases for 5 am (27%, -4.5 ppb, 95% CI: -5.7, -3.3) through 7 am (24%, -4.0 ppb, 95% CI: -5.2, -2.8), followed by 6 pm and 7 pm (22%, -3.7 ppb, 95% CI: -4.8, -2.6 and 22%, -4.8, -2.5, respectively), while the smallest decreases occurred at 11 pm and 1 am (both: 11%, -1.9 ppb, 95% CI: -3.1, -0.7).

Conclusion

NY on Pause's impact on TRAP varied greatly diurnally. Decreases during early morning and evening time periods are likely due to decreases in traffic. Our results may be useful for planning traffic policies that vary by time of day, such as congestion tolling policies.

Keywords: Nitrogen dioxide, Traffic-related air pollution, Pandemic, Policy impacts, New York City

Graphical abstract

Image, graphical abstract

Open in a new tab

Introduction

Policy responses to the COVID-19 pandemic, such as large-scale stay-at-home orders, were implemented across the United States (US) and globally in 2020 (Brodeur et al., 2021). The pandemic and its response policies substantially disrupted daily life, with major impacts on the labor market (Brodeur et al., 2021), economies (Brodeur et al., 2021), and transportation systems (Sun et al., 2021, Yasin et al., 2021), among other systems. In New York City (NYC), the most populous city in the US and the site of an early COVID-19 outbreak (The New York Times 2021), stay-at-home orders were implemented on March 22, 2020 (Governor Cuomo Signs March 20 2020). Called NY on Pause, the policy closed all non-essential businesses, prohibited non-essential gatherings of any size, advised individuals to stay home, and required social distancing (Governor Cuomo Signs March 20 2020).

Numerous studies have shown the impact of stay-at-home orders on traffic and traffic-related air pollutants (TRAP), both in the US and globally (Yasin et al., 2021, Burns et al., 1994, Dey et al., 2021, Bauwens et al., 2020, Dutheil et al., 2020, Hudda et al., 2020, Li et al., 2020, Patel et al., 2020, 13, Siddiqui et al., 2020, Tobías et al., 2020, Pitiranggon et al., 2022, Fu et al., 2020, Bian et al., 2021, Elshorbany et al., 2021). Traffic decreased dramatically in many regions of the world (Yasin et al., 2021, Schuman, 2020, Bian et al., 2021). Pollutants that are emitted directly by traffic (primary pollutants), such as black carbon (Hudda et al., 2020, Patel et al., 2020, Tobías et al., 2020), NO2 (Burns et al., 1994, Dey et al., 2021, Bauwens et al., 2020, Dutheil et al., 2020, Li et al., 2020, Patel et al., 2020, Siddiqui et al., 2020, Tobías et al., 2020, Pitiranggon et al., 2022, Fu et al., 2020, Elshorbany et al., 2021), and particulate matter (PM) (Hudda et al., 2020, Li et al., 2020, Patel et al., 2020, Tobías et al., 2020, Pitiranggon et al., 2022, Fu et al., 2020), tended to decrease with traffic. These findings have been more consistent for NO2 (Dey et al., 2021, Tobías et al., 2020, Fu et al., 2020), an established tracer for TRAP (Beckerman et al., 2008) that tends to correlate strongly with traffic (Kendrick et al., 2015). However, not all air pollutants decreased during stay-at-home orders; studies have shown that ozone, a secondary pollutant formed from the chemical reaction of volatile organic compounds (VOCs), NOx (NO2 and NO), and sunlight (Jhun et al., 2015), generally increased (Li et al., 2020, Patel et al., 2020, Tobías et al., 2020, Fu et al., 2020).

Understanding decreases in primary TRAP is important, considering the negative health effects associated with TRAP. For example, epidemiologic studies have connected different primary traffic pollutants with increased risk of stroke (Shah et al., 2015, Yorifuji et al., 2013), incident myocardial infarction (Hart et al., 2013, Nawrot et al., 2011), childhood asthma development (Khreis et al., 2017), asthma exacerbation (Lai et al., 2018), certain cancers (Cohen et al., 2018), pregnancy loss (Kioumourtzoglou et al., 2019), heart failure (Carey et al., 2016), all-cause mortality (Hart et al., 2013, Hoek et al., 2013), and decreases in cognition (de Prado Bert et al., 2018). In NYC, TRAP contributes to an estimated 320 deaths and 870 hospitalizations from traffic-related particulate matter ≤ 2.5 μm (PM2.5) every year (Kheirbek et al., 2016).

Considering the negative impact of primary traffic pollutants on health, studying changes in air pollution as a result of stay-at-home orders is important not only in the context of understanding the effects of policy responses to the pandemic but also in projecting the impact of future large-scale traffic interventions in cities. For example, NYC is in the process of implementing a congestion pricing policy, which would charge drivers a toll to enter part of Manhattan (Gold, September 29, 2021). Various congestion pricing schemes have been implemented and proposed in cities globally, but NYC is the first US city to implement such a policy (Lehe, 2019). One main goal of the policy is to decrease traffic congestion; if successful, we could expect corresponding decreases in primary traffic pollutants, such as NO2 (Gold, September 29, 2021, United States Department of Transportation 2022, Metro Transit Authority 2022). Understanding decreases in traffic and NO2 during the pandemic is helpful in projecting impacts of a congestion pricing policy or other traffic interventions (see, e.g., Perera et al., 2020) that may cause a similar change in traffic patterns. An effective congestion pricing policy (either in NYC or in other cities) may both decrease traffic overall and attenuate rush hour peak traffic, especially if tolls are higher during rush hours as is the case in other cities such as Stockholm (Lehe, 2019). This change in traffic patterns would be similar to what we have previously shown occurred in NYC during NY on Pause (Shearston et al., 2021), and thus we might expect similar changes in NO2.

Previous studies in NYC have shown that traffic decreased during NY on Pause (Pitiranggon et al., 2022, Bian et al., 2021, Shearston et al., 2021), and correspondingly, primary traffic pollutants also decreased (Pitiranggon et al., 2022, Fu et al., 2020, Zangari et al., 2020, Shehzad et al., 2021). However, none of the air pollution studies describes diurnal changes in NO2, only two investigate differences by location, and none separate changes at roadside monitors from changes at non-roadside monitors. We contribute to this body of knowledge by addressing each of these research gaps. The objectives of this analysis are to assess: (1) the overall impact of NY on Pause on NO2 concentrations in NYC, and (2) effect modification by (a) weekday/weekend, (b) roadside versus non-roadside monitors, (c) time-of-day, and (d) geographic location. We hypothesized that (1) NO2 would decrease overall on the order of ∼25%, given that traffic decreased by approximately 48% (Bian et al., 2021) and traffic is not the only source of NO2, (2a) decreases would be greater for weekdays rather than weekends given the larger pandemic-related traffic declines during weekdays (Shearston et al., 2021), (2b) roadside monitors would have greater decreases considering their proximity to traffic sources and the short NO2 decay distance (∼ 300 m) (Beckerman et al., 2008), (2c) greater decreases would be seen for NO2 in morning hours when NO2 has been found to have stronger correlations with traffic (Kendrick et al., 2015), and (2d) variation by geographic location would be observed, given that monitors represent very different traffic areas in NYC.

Methods

Study design

We used an interrupted time series (ITS) study design (Bernal et al., 2017) to evaluate the impact of NY's stay-at-home order, “NY on Pause,” on TRAP, using data from January 1, 2018 through June 8, 2020. Briefly, in the ITS study design, an underlying trend in the given time series is identified, and the degree to which an intervention or policy “interrupts” the existing time trend is evaluated (Bernal et al., 2017). In this design, there is no control group, as the continuation or interruption of a pre-existing trend is instead evaluated. This study design was selected given that the intervention/policy being evaluated occurred at the population level, and so an appropriate control group does not exist. As the intervention period, we used the date NY on Pause went into effect (March 22, 2020) through the date the policy ended for NYC (June 8, 2020), when the city entered Phase 1 of the state's reopening plan (Gold and Stevens, 2020). NO2, a primary pollutant emitted from vehicles and a well-known marker of vehicular traffic due to its moderate to high correlation with multiple traffic-related pollutants, was used to represent TRAP (Beckerman et al., 2008).

Data sources and definitions

NO2 Data

Hourly NO2 concentrations were obtained from the Environmental Protection Agency (EPA) Air Quality System (AQS) (United States Environmental Protection Agency 2020). We used data from all six hourly monitors maintained by the New York State Department of Environmental Conservation and New Jersey State Department of Environmental Protection in the NYC region (Fig. 1 ) to represent NYC. These stations collect data using continuous monitors, and report hourly averages to the EPA AQS following designated reference and equivalent methods used for air quality regulation (Table 1 ). Raw data accessed from the EPA AQS API was used in this analysis. No data transformation was applied. Monitor locations included two near-road sites (Queens College Near Road, Fort Lee Near Road) to capture emissions from roads. Two locations in New Jersey (NJ) were included as they are near major entrances to Manhattan from NJ, and were therefore likely impacted by NY on Pause.

Fig. 1.

Fig 1

Open in a new tab

Map of the six hourly NO2 AQS monitors in the NYC area (red circles). Black lines outline the five boroughs of NYC, overlaid on a Google roadmap (Map data © 2022 Google).

Table 1.

EPA AQS monitors included in the analysis.

Monitor Name Monitor ID EPA Measurement Scale Description Method Reference IDa
Pfizer Lab Site 36-005-0133 Urban scale On campus of botanical garden EQNA-0514-212
IS 52 36-005-0110 Urban scale In the South Bronx neighborhood, an area with greater traffic RFNA-1289-074
Queens College Near Road 36-081-0125 Middle scale At Queens College campus near Long Island Expy EQNA-0512-200
Queens College 2 36-081-0124 Neighborhood At Queens College campus RFNA-1289-074
Jersey City 34-017-1002 Neighborhood In Journal Square neighborhood, inland from the Hudson River RFNA-1194-099
Fort Lee Near Road 34-003-0010 Microscale At entrance to George Washington bridge; major entrance to Manhattan from New Jersey RFNA-1289-074
Open in a new tab

Meteorological data

Hourly weather information was obtained from the National Aeronautics and Space Administration (NASA) North American Land Data Assimilation System's primary forcing data (Xia). Meteorological data included: total precipitation, short-wave radiation flux downwards, surface pressure, specific humidity, temperature, and zonal and meridional wind vectors. Wind data were then converted to continuous wind speed and an eight-category meteorological wind direction variable (N, NE, E, SE, S, SW, W, NW). Meteorological data were included in statistical models as potential confounders of the relationship between NY on Pause and NO2 concentrations because they vary with time in a manner similar to traffic and directly impact NO2 concentrations through atmospheric dispersion and influence on NOx chemical reactivity.

Long-term trends

To account for long-term (greater than 3 years) and seasonal trends in NO2 concentrations not fully captured by the ITS design, models were additionally adjusted for year and month as categorical variables.

Statistical analyses

First, descriptive statistics for all variables were conducted. Next, we constructed a linear mixed effects model with random intercepts for each monitor, serving as the main model. While the distribution of NO2 was mildly right skewed, we chose not to scale or log transform NO2 or meteorological covariates in order to facilitate interpretation of the results for policymakers. We modeled all variables linearly and included all meteorological variables in the model regardless of their statistical significance. Additionally, we included year and month to account for long-term and seasonal trends, and time of day (a 24-level categorical variable) and day of week (a 7-level categorical variable) to account for potential confounding by other NO2 sources that covaried with traffic. We assumed that NY on Pause would cause an immediate and sharp decrease in NO2 concentrations and used the ITS study design to model a step change in NO2 concentrations. We therefore constructed two additional variables: a binary variable to represent the intervention, NY on Pause, coded as 0 from January 1, 2018 through March 21, 2020, and 1 from March 22, 2020 through June 8, 2020; and a continuous variable representing time elapsed since the beginning of the analysis (January 1, 2018).

To assess effect modification by weekend/weekday status, we ran stratified models by weekend and weekday, which were otherwise the same as described above. Similarly, to assess differences in NO2 concentrations measured at roadside vs. non-roadside monitors, we stratified by roadside vs non-roadside status. To assess diurnal changes in NO2 concentrations, we ran stratified models by time of day. Short-wave radiation flux downwards was removed from the models for 8 pm to 4 am because it was always 0 W/m2, but the model was otherwise the same. Effect estimates from stratified models were compared using the Wald test and Cochran test to determine if estimates were significantly different, using a p-value cutoff of 0.05 (Kaufman and MacLehose, 2013). To evaluate potential variation in effect estimates by monitor location, we added a random slope for monitor ID to the main model, and then compared Akaike information criterion (AIC)s between the two models (random intercepts only vs. random intercepts and slopes) and ran a likelihood ratio test to determine if the models were significantly different at p = 0.05.

We present model results as percent decreases in pre-NY on Pause NO2 concentrations, calculated by dividing the effect estimate by the mean NO2 concentration from January 1, 2018 through March 21, 2020 and multiplying by 100%.

Sensitivity analyses

To confirm the stability of our results, several sensitivity analyses were performed. First, in the subset of observations for which we had PM2.5 data (n = 86,093; 70% of observations in the main model), we added hourly non-traffic PM2.5 concentrations to the main model to account for potential confounding by other air pollution sources that may covary with traffic and thus traffic-related NO2 (e.g., industries that emit both PM2.5 and NO2). To implement this adjustment, we regressed PM2.5 on NO2 (to isolate traffic-related PM2.5) and added the residuals (non-traffic-related PM2.5) to the main model as a covariate. Second, we imputed missing NO2 observations (n = 5,708; 4.5%) using the mean NO2 value for the year, season, time of day, and day of week of each missing observation, and repeated the main analysis. Third, we removed potential outliers (defined as any observations with a model residual value greater than 3 standard deviations from the mean) and repeated the main analysis. Fourth, we repeated the main analysis using March 20 as the start date for NY on Pause, as this was when the policy was announced by the governor (Governor Cuomo Signs, 2020), prompting a decrease in traffic two days before policy implementation (Bian et al., 2021). Fifth, we removed the two monitors from New Jersey and repeated the main analysis, as New Jersey was subject to its own stay-at-home order (Governor Murphy Annoucnes Statewide Stay at Home Order March 21 2020) which may have influenced NO2 concentrations at those monitors. Finally, we conducted detrending and meteorological normalization to account for short- and long-term trends and meteorology rather than adjusting for these variables in the mixed effects model, and repeated the main analysis.

All analyses were conducted in R version 4.1.1 R Core Team, 2021. The nlme package (nlme 2021) was used to run the mixed effect models. Code to recreate all analyses in this manuscript can be found here: https://github.com/jenni-shearston/NYonPAUSE-NO2-ITS

Results

NO2 concentrations from January 1, 2018 through June 8, 2020 followed the expected seasonal pattern, with increased concentrations over the winter months and decreased concentrations over the summer months (Fig. 2 ). Descriptive statistics for NO2 concentrations and weather parameters before and during NY on Pause are shown in Table 2 . Briefly, unadjusted NO2 concentrations were lower during NY on Pause (mean = 10.9 ppb; standard deviation [sd] = 7.4 ppb) as compared to the period before (mean = 16.8 ppb; sd = 10.5 ppb). Weather parameters remained similar across the two periods, with the exception of short-wave radiation flux downwards, which—given the absence of darker winter months during NY on Pause—was expected to be higher during stay-at-home orders. Additionally, wind roses show some difference in wind patterns between the pre-NY on Pause and NY on Pause periods (Supplemental Fig. 1). A histogram of the NO2 concentrations showed a mildly right skewed distribution (Supplemental Fig. 3).

Fig. 2.

Fig 2

Open in a new tab

Time series plot of daily NO2 concentrations from January 1, 2018 through June 8, 2020. Daily rather than hourly concentrations were used for ease in visualization. The dashed, vertical line indicates the beginning of the stay-at-home order, NY on Pause, on March 22, 2020. Colors indicate the six monitors in the NYC area, and solid-colored lines represent the smoothed average for the entire study period.

Table 2.

Pre vs. during intervention NO2 concentrations and weather parameters.

Parameter Pre-intervention
 During-intervention
Mean SD Mean SD
NO2 (ppb) 16.8 10.5 10.9 7.4
Hourly precipitation (mm/hr) 0.12 0.56 0.09 0.41
Short-wave radiation flux downwards (W/m2) 178.0 255.6 235.3 292.5
Specific humidity (g/kg) 7.9 5.0 6.8 3.0
Surface pressure (kPa) 101.5 0.8 101.2 0.8
Temperature (°C) 11.7 9.9 11.8 5.6
Wind speed (m/s) 5.2 2.7 5.2 2.6
Open in a new tab

Using the fully adjusted ITS model, we found that NY on Pause was followed by a significant 19% decrease in pre-NY on Pause NO2 concentrations (-3.2 ppb, 95% confidence interval [CI]: -3.5, -3.0) (Table 3 ). Model diagnostic plots can be found in the supplemental material (Supplemental Figs. 4-7) and in the R script accompanying this manuscript (2_02_models_no2-its_manuscript, Section 2). Decreases during weekend days vs weekday days were very similar to the overall effect and had insignificant Wald and Cochran test statistics, indicating no effect modification. In contrast, we found evidence for effect modification by roadside vs non-roadside monitors, where non-roadside monitors had larger decreases in NO2 concentrations compared to roadside monitors (non-roadside = 20.2% [-3.4 ppb], 95% CI: -3.7, -3.1; roadside = 16.1% [-2.7], 95% CI: -3.1, -2.3; p-value < 0.05).

Table 3.

Effect of NY on Pause on hourly NO2 concentrations (ppb) in NYC.

Percent Decreasec Effect Estimate 95% Confidence Interval
Overall effecta 19.0% -3.2 -3.5, -3.0
Weekdaysb 19.0% -3.2 -3.5, -2.9
Weekendsb 19.0% -3.5 -3.9, -3.0
Roadside monitors onlya 16.1% -2.7 -3.1, -2.3
Non-roadside monitors onlya 20.2% -3.4 -3.7, -3.1
Open in a new tab
a

Adjusted for day of week, time of day, month, year, wind direction, wind speed, temperature, precipitation, downward flux solar radiation, specific humidity, and surface pressure

b

Not adjusted for day of week

c

Percentage of pre-NY on Pause NO2 concentrations

NY on Pause had differential impacts on NO2 concentrations depending on the hour of day, although there were significant decreases in NO2 for all hours (Fig. 3 ). The largest decreases were on the order of ∼ 22 to 27% and occurred for 5 am (26.8% or -4.5 ppb, 95% CI: -5.7, -3.3), 6 am (27.4% or -4.6 ppb, 95% CI: -5.9, -3.4), and 7 am (23.8% or -4.0 ppb, 95% CI: -5.2, -2.8) followed by 6 pm (22.0% or -3.7 ppb, 95% CI: -4.8, -2.6) and 7 pm (22.0% or -3.7 ppb, 95% CI: -4.8, -2.5). The smallest decreases occurred for the hours of 11pm and 1am (both: 11.3% or -1.9 ppb, 95% CI: -3.1, -0.7). Again, both Wald and Cochran tests were significant (p < 0.05), providing further evidence of effect modification by hour of day.

Fig. 3.

Fig 3

Open in a new tab

Percent decrease in pre-NY on Pause concentrations, effect estimates, and 95% confidence intervals describing the impact of NY on Pause on NO2 (ppb) concentrations in NYC, stratified by hour. Models were adjusted for day of week, month, year, wind direction, wind speed, temperature, precipitation, specific humidity, and surface pressure. Models for 5 am to 7 pm were additionally adjusted for downward flux solar radiation.

We found weak evidence for modification by monitor location, as adding random slopes to the main model caused the AIC to decrease very slightly (although significantly; p = 0.0026), with random slopes clustering near the fixed effect (Supplemental Fig. 2).

Results from sensitivity analyses were similar to the main results. In the subset of data with available PM2.5 concentrations, adding PM2.5 to the model resulted in a slightly attenuated main effect (a 17.9% decrease or -3.0 ppb, 95% CI: -3.2, -2.7). While this effect estimate is not directly comparable to the main analysis, results suggest that PM2.5 had minimal impact on our results. Imputing missing NO2 concentrations, removing NO2 concentrations greater than 3 standard deviations from the mean, changing the start date of the intervention to March 20 rather than March 22, removing NJ monitors, and conducting detrending and meteorological normalization all resulted in findings very similar to those of the main model (results not shown).

Overall, we found that NY on Pause had a substantial impact on NO2 concentrations, resulting in a larger Spring seasonal decrease than would otherwise be expected. In Fig. 4 , predicted NO2 concentrations for both the counterfactual scenario where NY on Pause was not enacted and the actual scenario are shown, depicting the lower-than-expected NO2 concentrations during the NY on Pause period.

Fig. 4.

Fig 4

Open in a new tab

Time series of daily mean NO2 concentrations (dots) predicted from our mixed effects model and 14-day rolling averages (lines) for the actual NY on Pause scenario (blue) and the hypothetical counterfactual scenario where NY on Pause was not enacted (purple). The dashed, vertical line indicates the beginning of the intervention, NY on Pause, on March 22, 2020.

Discussion

NY on Pause caused an average decrease in NO2 concentrations of 3.2 ppb (95% CI: -3.5, -3.0), or 19% of pre-intervention concentrations. However, this varied dramatically throughout the day, ranging from decreases of 27% to 11%, with the largest decreases observed for 5 to 7 am and 6 to 7 pm. This overlapped with the morning and evening rush hour travel peaks, which in NYC range from approximately 6 to 10 am in the morning and 4 to 8 pm in the evening (The Staten Island Ferry 2022, NJ Transit. Schedule Information. 2021, The Port Authority of New York and New Jersey 2022, Metro Transit Authority 2022). The smallest decreases occurred overnight at 11 pm and 1 am. The diurnal pattern of these decreases (larger during rush hour periods) suggests that commuter traffic, rather than trucks or freight traffic, may have been most impacted by the policy.

Our findings corroborate other studies evaluating the impact of the COVID-19 pandemic and stay-at-home orders on traffic-related air pollution in NYC. Using data from the New York City Community Air Survey, Pitiranggon et al., 2022 found a decrease in NO2 of 29%, with larger changes occurring in the central business district area of Manhattan (Pitiranggon et al., 2022). While our study found a smaller decrease, Pitiranggon et al. used data with high spatial resolution and coverage in the central business district, an area that is not close to any of the air pollution monitors maintained by the state. This may explain our lower estimates and suggests that our findings may be biased towards the null. Fu et al (Fu et al., 2020) and Elshorbany et al (Elshorbany et al., 2021) also found decreases in NO2 during stay-at-home orders, using a daily measure of NO2 Air Quality Index and daily OMI satellite measurements of the NO2 tropospheric column, respectively. They found decreases ranging from 33.7 to 50% in the NYC area. As each study used a different data source and slightly different comparison period, percentages are not directly comparable between studies; however, their consistency in direction and magnitude are reassuring. Of note, Zangari et al., 2020 did not find a significant impact of stay-at-home orders on NO2 concentrations using a linear time lag model (Zangari et al., 2020). However, the authors compared linear trends in NO2 concentrations for January through May for several calendar years, treating 2020 as the intervention year, rather than assessing the impact of the stay-at-home order itself. Additionally, there were many more pre-intervention NO2 observations included in the time series for 2020 than during-intervention observations, likely biasing the trend towards pre-intervention concentrations.

Additionally, several studies have found that traffic, the primary source of NO2, also decreased dramatically as a result of NY on Pause (Pitiranggon et al., 2022, Bian et al., 2021, Shearston et al., 2021). Our previous analysis of congestion levels in Manhattan found not only an overall decrease in congestion but also a large depression of the morning and evening rush hour peaks (Shearston et al., 2021). Similarly, Pitiranggon et al., 2022 found decreases in daytime traffic speeds across NYC, with large decreases in travel times during morning rush hour at the Lincoln Tunnel, a major entrance to Manhattan and its central business district (Pitiranggon et al., 2022). Both studies found smaller decreases for overnight times (Pitiranggon et al., 2022, Shearston et al., 2021). These findings support the results of our time-of-day analysis; as the largest decreases in traffic occur during the day and during rush hour peaks, it makes sense that NO2, as a tracer of TRAP and primary traffic pollutant, would also follow a similar pattern. This is also consistent with a pre-pandemic study of diurnal NO2 in New York, which also described peaks in NO2 during similar times of day– in the winter and spring from approximately 6 to 9 am, in conjunction with a similarly timed peak in NO, and in the evening from 3pm to midnight (with a broad and low NO peak) (Civerolo et al., 2017). This diurnal pattern resulted from a combination of increased primary emissions during rush hour, atmospheric chemistry as NOx begins reacting with sunlight to form ozone later in the day, and a lower planetary boundary layer in the mornings and evenings (Civerolo et al., 2017). Thus, it makes sense that we saw the largest decreases in NO2 during the same rush hour periods, not only due to decreases in traffic emissions during these peak time periods, but also because the lower planetary boundary layer and weaker sunlight in early morning and evening hours might amplify decreases compared to midday when the boundary layer is higher (diluting impacts) (Jhun et al., 2015).

While we found weak evidence for variation in NO2 reductions by location, we were limited by the six monitors included in the study. These monitors represented varying types of urban areas and spatial scales, with some being located right next to major roads and some in greener areas further from roads, however, using only 6 monitors may not adequately capture the heterogeneity in NO2 concentrations. For example, Pitiranggon et al., 2022 found substantial geographic variation within NYC, using highly spatially resolved data (Pitiranggon et al., 2022). Nonetheless, we were able to detect that roadside monitors had smaller decreases in NO2 than non-roadside monitors. The two roadside monitors were located on major interstate highways (I95 at the George Washington Bridge and I296 at Queens College) and so may have had proportionally more freight and delivery traffic, as well as overall a smaller reduction in traffic, considering their nature as traffic arteries through the city.

The findings of our study have implications for future large-scale traffic policies that may be implemented in NYC. For example, a congestion pricing plan known as the Central Business District Tolling Program has been approved by New York State government and is in the early process of implementation (Gold, 2021). As with many such policies, a major potential benefit of the program is the decrease in traffic-related air pollution that may occur with a successful decrease in congestion and overall traffic. Our analysis of NO2 changes in response to NY on Pause can be used to inform the design of tolling schemes for the congestion pricing policy and others. Following dramatic decreases in traffic congestion after NY on Pause, and especially decreases during peak rush hour time points, our current study correspondingly found the largest decreases in NO2 during early morning and evening time periods, and the smallest decreases overnight. Larger decreases during rush hour peaks likely suggest that NY on Pause had a greater impact on commuter vehicles than trucks or freight vehicles. This is supported by vehicle counts from Port Authority bridge and tunnel crossings in NYC (Supplemental Table 1). In January and February of 2020, automobiles made up 91% of all crossings, while trucks made up 6-7%. In April, while counts of all types of vehicles decreased dramatically, the proportion of automobile to freight traffic also shifted: 87% of all crossings were automobiles, and 11% were trucks. These patterns were slightly stronger at the bridge and tunnel site that also included an NO2 monitor (Fort Lee near road monitor): the George Washington Bridge. It is likely that the Congestion Pricing policy will also have a greater impact on commuter vehicles than freight trucks, as commuters can switch from private vehicles to public transit, while trucks/freight will still need to make deliveries.

While reductions in NO2 occurred, it should be noted that several studies have found that ozone concentrations increased during stay-at-home orders (Li et al., 2020, Patel et al., 2020, Tobías et al., 2020, Fu et al., 2020), likely due to decreases in NOx from traffic reductions, as NO scavenges ozone (especially during nighttime hours), leading to decreased ozone concentrations (Jhun et al., 2015). With a reduction in NOx primary emissions from traffic, we may expect ozone to increase, especially if VOC concentrations do not decrease at similar magnitudes, as VOCs react with NOx and sunlight to form ozone. This may be the case in NYC, where volatile consumer and industrial products also contribute substantially to VOCs (in addition to vehicular traffic) (Coggon et al., 2021). Accordingly, any traffic policy should also take non-linear ozone chemistry and VOC source contributions into consideration. To achieve NO2 decreases in the range we report from NY on Pause with congestion reducing policies, this supports implementation of a tolling scheme that charges higher prices during early morning and evening time periods, aiming to both decrease total traffic and spread the overall amount of traffic throughout the day. Such variable tolling schemes have precedent in other major cities, including London and Stockholm, and are a common component of downtown congestion pricing schemes (Lehe, 2019).

In addition to implications for traffic policies in NYC, our study has implications for potential health interventions. Cardiovascular events like stroke and myocardial infarction can be triggered by traffic-related air pollution (Shah et al., 2015, Yorifuji et al., 2013, Hart et al., 2013, Nawrot et al., 2011). Understanding diurnal changes in NO2 concentrations in NYC can thus lead to recommendations for people at risk of these events (and others) about when to partake in outdoor activities. However, changes in ozone and other pollutants should also be considered in these decisions, as they too are associated with adverse health outcomes (Soares and Silva, 2022, Guo et al., 2022, Liu et al., 2022).

Our study has several strengths and limitations. First, we had access to a clear policy implementation time point and adjusted for numerous covariates, such as meteorology and long-term and seasonal trends, which could have confounded the relationship between NY on Pause and NO2 concentrations. As with other ITS designs, we cannot rule out the possibility that other factors led to the decrease in NO2 concentrations, however, given the clear directive of the stay-at-home order and evidence from previous studies that NY on Pause in isolation dramatically impacted traffic (Bian et al., 2021), it is likely that the effect we found was most influenced by NY on Pause. However, there is still the possibility of residual confounding from other variables, especially those pertinent to the fate of atmospheric NO2 (such as VOCs), and bias from specification of the start date of the NY on Pause intervention. To increase confidence in our results, we conducted several sensitivity analyses, finding our model specification robust to confounding by particulate matter, specification of the start date of the intervention, missingness of NO2 concentrations, and exclusion of outlier NO2 values. Second, we included NO2 monitors from New Jersey in this analysis. Considering that New Jersey implemented its own version of stay-at-home orders, (Governor Murphy Annoucnes Statewide Stay at Home Order, 2020), NO2 values from these stations likely represented traffic decreases caused by both NY on Pause and by New Jersey policies. Our sensitivity analysis removing New Jersey monitors supports this conclusion, as removing New Jersey monitors decreased the effect estimate by a modest 0.05 ppb. However, this impact was small. Third, we used diurnally resolved air pollution data, which allowed us to assess how NY on Pause impacted traffic-related air pollution at the hourly level and has important policy implications. However, our data had poor spatial resolution, relying upon only six monitors to represent the NYC area. This prevented us from making inferences about variation within NYC, and also may have biased our results toward the null, as we did not have a monitor located in one of the most traffic intensive areas of NYC, the central business district. Of particular concern, we only had two near-road sites. Importantly, poor spatial resolution prevented us from assessing disparities in air pollution reductions as a result of NY on Pause. We echo the calls of other researchers and advocates to predict and evaluate the impact of NY on Pause and future policies like congestion pricing through an environmental justice lens, ensuring that the benefits and costs of new policies are equitably shared. Such work will necessitate the collection and use of highly spatially resolved data.

Conclusion

New York's stay-at-home order (NY on Pause) implemented in response to the COVID-19 pandemic decreased NO2 concentrations by an average of 3.2 ppb (19%), with substantial diurnal variation. The greatest decreases in NO2 concentrations were seen from 5 to 7 am and 6 to 7 pm, and lowest decreases for 11 pm and 1 am. Our findings could suggest that other large-scale policies that aim to reduce traffic such as NYC's planned congestion pricing policy may result in substantial NO2 decreases; however, changes in other pollutants such as ozone should also be considered to assess overall improvements in air quality. Paired with highly spatially resolved air pollution data, our results could be used to investigate equity in implementation of NY on Pause and other future traffic and traffic-related air pollution policies.

CRediT authorship contribution statement

Jenni A. Shearston: Conceptualization, Data curation, Formal analysis, Methodology, Visualization, Writing – original draft. Ilan Cerna-Turoff: Validation, Writing – review & editing. Markus Hilpert: Conceptualization, Supervision, Writing – review & editing. Marianthi-Anna Kioumourtzoglou: Conceptualization, Methodology, Supervision, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Funding

JAS was supported by grants from the National Institute of Environmental Health Sciences (NIEHS grant # F31 ES033098 and grant # 2T32 ES7322-19). This work was also partially supported by NIEHS P30 ES009089, R21 ES030093, and R01 ES030616.

Footnotes

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.heha.2022.100032.

Appendix. Supplementary materials

mmc1.docx (20.9MB, docx)

References

  1. Bauwens M, Compernolle S, Stavrakou T, et al. Impact of coronavirus outbreak on NO2 pollution assessed using TROPOMI and OMI observations. Geophys. Res. Lett. 2020;47(11) doi: 10.1029/2020GL087978. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Beckerman B, Jerrett M, Brook JR, Verma DK, Arain MA, Finkelstein MM. Correlation of nitrogen dioxide with other traffic pollutants near a major expressway. Atmos. Environ. 2008;42(2):275–290. [Google Scholar]
  3. Bernal JL, Cummins S, Gasparrini A. Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int. J. Epidemiol. 2017;46(1):348–355. doi: 10.1093/ije/dyw098. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Bian Z, Zuo F, Gao J, et al. Time lag effects of COVID-19 policies on transportation systems: a comparative study of New York City and Seattle. Transp. Res. Part A. 2021;145:269–283. doi: 10.1016/j.tra.2021.01.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Brodeur A, Gray D, Islam A, Bhuiyan S. A literature review of the economics of COVID-19. J. Econ. Surv. 2021;35(4):1007–1044. doi: 10.1111/joes.12423. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Burns J, Hoffmann S, Kurz C, Laxy M, Polus S, Rehfuess E. COVID-19 mitigation measures and nitrogen dioxide – A quasi-experimental study of air quality in Munich, Germany. Atmos. Environ. 1994;2020 doi: 10.1016/j.atmosenv.2020.118089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Carey IM, Anderson HR, Atkinson RW, et al. Traffic pollution and the incidence of cardiorespiratory outcomes in an adult cohort in London. Occup. Environ. Med. 2016;73(12):849–856. doi: 10.1136/oemed-2015-103531. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Civerolo KL, Rattigan OV, Felton HD, Schwab JJ. Changes in gas-phase air pollutants across New York State, USA. Aerosol. Air Qual. Res. 2017;17(1):147–166. [Google Scholar]
  9. Coggon MM, Gkatzelis GI, McDonald BC, et al. Volatile chemical product emissions enhance ozone and modulate urban chemistry. Proc. Natl. Acad. Sci. 2021;118(32) doi: 10.1073/pnas.2026653118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Cohen G, Levy I, Yuval, et al. Chronic exposure to traffic-related air pollution and cancer incidence among 10,000 patients undergoing percutaneous coronary interventions: a historical prospective study. Eur. J. Prev. Cardiol. 2018;25(6):659–670. doi: 10.1177/2047487318760892. [DOI] [PubMed] [Google Scholar]
  11. de Prado Bert P, Mercader EMH, Pujol J, Sunyer J, Mortamais M. The effects of air pollution on the brain: a review of studies interfacing environmental epidemiology and neuroimaging. Curr. Environ. Health Rep. 2018;5(3):351–364. doi: 10.1007/s40572-018-0209-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Dey T, Tyagi P, Sabath MB, et al. Counterfactual time series analysis of short-term change in air pollution following the COVID-19 state of emergency in the United States. Sci. Rep. 2021;11(1):23517. doi: 10.1038/s41598-021-02776-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dutheil F, Baker JS, Navel V. COVID-19 as a factor influencing air pollution? Environ. Pollut. 2020;263 doi: 10.1016/j.envpol.2020.114466. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Elshorbany YF, Kapper HC, Ziemke JR, Parr SA. The status of air quality in the United States during the COVID-19 pandemic: a remote sensing perspective. Remote Sens. 2021;13(3):369. [Google Scholar]
  15. Fu F, Purvis-Roberts KL, Williams B. Impact of the COVID-19 Pandemic lockdown on air pollution in 20 major cities around the world. Atmosphere. 2020;11(11):1189. [Google Scholar]
  16. Gold M, Stevens M. 2020. What are the Phases of New York's Reopening Plan? The New York TImes. [Google Scholar]
  17. Gold M. The New York Times; 2021. Congestion Pricing Is Coming to New York. Everyone Has an Opinion. [Google Scholar]
  18. Governor Cuomo Signs the “New York State on PAUSE” Executive Order [press release]. New York State, March 20 2020.
  19. Governor Murphy Annoucnes Statewide Stay at Home Order, Closure of All Non-Essential Retail Businesses [press release]. Newark, NJ: New Jersey State, March 21 2020.
  20. Guo X, Su W, Wang H, et al. Short-term exposure to ambient ozone and cardiovascular mortality in China: a systematic review and meta-analysis. Int. J. Environ. Health Res. 2022:1–18. doi: 10.1080/09603123.2022.2066070. [DOI] [PubMed] [Google Scholar]
  21. Hart JE, Rimm EB, Rexrode KM, Laden F. Changes in traffic exposure and the risk of incident myocardial infarction and all-cause mortality. Epidemiology. 2013;24(5):734–742. doi: 10.1097/EDE.0b013e31829d5dae. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hoek G, Krishnan RM, Beelen R, et al. Long-term air pollution exposure and cardio- respiratory mortality: a review. Environ. Health. 2013;12(1):43. doi: 10.1186/1476-069X-12-43. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Hudda N, Simon MC, Patton AP, Durant JL. Reductions in traffic-related black carbon and ultrafine particle number concentrations in an urban neighborhood during the COVID-19 pandemic. Sci. Total Environ. 2020;742 doi: 10.1016/j.scitotenv.2020.140931. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Jhun I, Coull BA, Zanobetti A, Koutrakis P. The impact of nitrogen oxides concentration decreases on ozone trends in the USA. Air Qual. Atmos. Health. 2015;8(3):283–292. doi: 10.1007/s11869-014-0279-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kaufman JS, MacLehose RF. Which of these things is not like the others? Cancer. 2013;119(24):4216–4222. doi: 10.1002/cncr.28359. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kendrick CM, Koonce P, George LA. Diurnal and seasonal variations of NO, NO2 and PM2.5 mass as a function of traffic volumes alongside an urban arterial. Atmos. Environ. 2015;122:133–141. [Google Scholar]
  27. Kheirbek I, Haney J, Douglas S, Ito K, Matte T. The contribution of motor vehicle emissions to ambient fine particulate matter public health impacts in New York City: a health burden assessment. Environ. Health. 2016;15(1):89. doi: 10.1186/s12940-016-0172-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Khreis H, Kelly C, Tate J, Parslow R, Lucas K, Nieuwenhuijsen M. Exposure to traffic-related air pollution and risk of development of childhood asthma: A systematic review and meta-analysis. Environ. Int. 2017;100:1–31. doi: 10.1016/j.envint.2016.11.012. [DOI] [PubMed] [Google Scholar]
  29. Kioumourtzoglou MA, Raz R, Wilson A, et al. Traffic-related air pollution and pregnancy loss. Epidemiology. 2019;30(1):4–10. doi: 10.1097/EDE.0000000000000918. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Lai VWY, Bowatte G, Knibbs LD, et al. Residential NO2 exposure is associated with urgent healthcare use in a thunderstorm asthma cohort. Asia Pac Allergy. 2018;8(4):e33. doi: 10.5415/apallergy.2018.8.e33. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Lehe L. Downtown congestion pricing in practice. Transp. Res. Part C. 2019;100:200–223. [Google Scholar]
  32. Li L, Li Q, Huang L, et al. Air quality changes during the COVID-19 lockdown over the Yangtze River Delta Region: An insight into the impact of human activity pattern changes on air pollution variation. Sci. Total Environ. 2020;732 doi: 10.1016/j.scitotenv.2020.139282. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Liu N, Bu Z, Liu W, et al. Health effects of exposure to indoor volatile organic compounds from 1980 to 2017: a systematic review and meta-analysis. Indoor Air. 2022;32(5):e13038. doi: 10.1111/ina.13038. [DOI] [PubMed] [Google Scholar]
  34. Metro Transit Authority. LIRR and Metro-North fares. https://new.mta.info/fares/lirr-metro-north. Accessed June 1, 2022.
  35. Metro Transit Authority . 2022. Central Business District Tolling Program.https://new.mta.info/project/CBDTP Accessed October 14, 2022. [Google Scholar]
  36. Nawrot TS, Perez L, Künzli N, Munters E, Nemery B. Public health importance of triggers of myocardial infarction: a comparative risk assessment. Lancet North Am. Ed. 2011;377(9767):732–740. doi: 10.1016/S0140-6736(10)62296-9. [DOI] [PubMed] [Google Scholar]
  37. NJ Transit . 2021. Schedule Information.https://www.njtransit.com/schedules Accessed June 1, 2022. [Google Scholar]
  38. nlme . 2021. Linear and Nonlinear Mixed Effect Models [computer program] R package version 3.1-152. [Google Scholar]
  39. Patel H, Talbot N, Salmond J, Dirks K, Xie S, Davy P. Implications for air quality management of changes in air quality during lockdown in Auckland (New Zealand) in response to the 2020 SARS-CoV-2 epidemic. Sci. Total Environ. 2020;746 doi: 10.1016/j.scitotenv.2020.141129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Perera F, Berberian A, Cooley D, et al. Health benefits of sustained air quality improvements in New York City: a simulation based on air pollution levels during the COVID-19 shutdown. Environ. Res. 2020 doi: 10.1016/j.envres.2020.110555. [DOI] [PubMed] [Google Scholar]
  41. Pitiranggon M, Johnson S, Huskey C, Eisl H, Ito K. Effects of the COVID-19 shutdown on spatial and temporal patterns of air pollution in New York City. Environ. Adv. 2022 [Google Scholar]
  42. R Core Team . R Foundation for Statistical Computing; Vienna, Austria: 2021. R: A Language and Environment for Statistical Computing [Computer Program] [Google Scholar]
  43. Schuman R. 2020. INRIX U.S. National Traffic Volume Synopsis: Issue #5 (April 11 –17, 2020). INRIX. [Google Scholar]
  44. Shah AS, Lee KK, McAllister DA, et al. Short term exposure to air pollution and stroke: systematic review and meta-analysis. BMJ. 2015;350:h1295. doi: 10.1136/bmj.h1295. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Shearston JA, Martinez ME, Nunez Y, Hilpert M. Social-distancing fatigue: evidence from real-time crowd-sourced traffic data. Sci. Total Environ. 2021 doi: 10.1016/j.scitotenv.2021.148336. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Shehzad K, Bilgili F, Koçak E, Xiaoxing L, Ahmad M. COVID-19 outbreak, lockdown, and air quality: fresh insights from New York City. Environ. Sci. Pollut. Res. Int. 2021;28(30):41149–41161. doi: 10.1007/s11356-021-13556-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Siddiqui A, Halder S, Chauhan P, Kumar P. COVID-19 pandemic and city-level nitrogen dioxide (NO 2) reduction for urban centres of India. J. Indian Soc. Remote Sens. 2020:1–8. [Google Scholar]
  48. Soares AR, Silva C. Review of ground-level ozone impact in respiratory health deterioration for the past two decades. Atmosphere. 2022;13(3):434. [Google Scholar]
  49. Sun X, Wandelt S, Zheng C, Zhang A. COVID-19 pandemic and air transportation: Successfully navigating the paper hurricane. J. Air Transp. Manag. 2021;94 doi: 10.1016/j.jairtraman.2021.102062. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. The New York Times . 2021. Tracking Coronavirus in New York: Latest Map and Case Count.https://www.nytimes.com/interactive/2021/us/new-york-covid-cases.html Accessed December 8, 2021. [Google Scholar]
  51. The Port Authority of New York and New Jersey . 2022. Tolls.https://www.panynj.gov/bridges-tunnels/en/tolls.html Accessed June 1, 2022. [Google Scholar]
  52. The Staten Island Ferry . 2022. Frequently Asked Questions.https://www.siferry.com/faqtips.html Accessed June 1, 2022. [Google Scholar]
  53. Tobías A, Carnerero C, Reche C, et al. Changes in air quality during the lockdown in Barcelona (Spain) one month into the SARS-CoV-2 epidemic. Sci. Total Environ. 2020;726 doi: 10.1016/j.scitotenv.2020.138540. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. United States Department of Transportation . 2022. Central Business District (CBD) Tolling Program Environmental Assessment. [Google Scholar]
  55. United States Environmental Protection Agency . 2020. Air Quality System (AQS) [Google Scholar]; https://www.epa.gov/aqs#:˜:text=The%20Air%20Quality%20System%20(AQS,from%20over%20thousands%20of%20monitors  Accessed July 12, 2020.
  56. Xia Y, et al. NLDAS Primary Forcing Data L4 Hourly 0.125 x 0.125 degree V002. Goddard Earth Sciences Data and Information Services Center (GES DISC). Dataset. 10.5067/6J5LHHOHZHN4
  57. Yasin YJ, Grivna M, Abu-Zidan FM. Global impact of COVID-19 pandemic on road traffic collisions. World J. Emerg. Surg. 2021;16(1):51. doi: 10.1186/s13017-021-00395-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Yorifuji T, Kashima S, Tsuda T, et al. Long-term exposure to traffic-related air pollution and the risk of death from hemorrhagic stroke and lung cancer in Shizuoka, Japan. Sci. Total Environ. 2013;443:397–402. doi: 10.1016/j.scitotenv.2012.10.088. [DOI] [PubMed] [Google Scholar]
  59. Zangari S, Hill DT, Charette AT, Mirowsky JE. Air quality changes in New York City during the COVID-19 pandemic. Sci. Total Environ. 2020;742 doi: 10.1016/j.scitotenv.2020.140496. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

mmc1.docx (20.9MB, docx)

Articles from Hygiene and Environmental Health Advances are provided here courtesy of Elsevier

RESOURCES