SARS-CoV-2 test positivity rate in Reno, Nevada: association with PM2.5 during the 2020 wildfire smoke events in the western United States

The study investigates whether elevated PM2.5 from 2020 western U.S. wildfire smoke increased SARS-CoV-2 infections in Reno, Nevada. During summer–fall 2020, widespread wildfires coincided with a COVID-19 surge, exposing residents to prolonged PM2.5. Prior evidence links air pollution, particularly PM2.5, to increased susceptibility to respiratory viruses and potentially to increased SARS-CoV-2 transmission and severity. Reno, situated in an intermountain valley with constrained pollutant dispersion and heavily impacted by wildfire smoke, provided a setting to examine the association between ambient PM2.5 and SARS-CoV-2 test positivity. The research question: Is wildfire-related PM2.5 associated with increased SARS-CoV-2 test positivity rates? The purpose is to quantify this association and estimate excess cases attributable to wildfire smoke to inform public health preparedness.

The paper situates its work within literature showing air pollution’s adverse health effects, especially PM2.5 as a key mediator of respiratory morbidity. Mechanistic studies suggest PM2.5 can modulate immune responses, increase airway/systemic inflammation, and act as a carrier enhancing bioaerosol spread, including SARS-CoV-2. Historical associations exist between particulate pollution and SARS mortality (SARS-CoV-1). Reviews have summarized links between PM/NO2 and COVID-19 spread/lethality. Wildfire-derived PM2.5 has documented respiratory impacts. Epidemiological studies have reported associations between air pollution and COVID-19 infectivity/severity in multiple countries, with mixed findings in Spain. During the 2020 wildfires, studies in San Francisco and Orange County reported increased COVID-19 cases with higher PM2.5, though with fewer covariate adjustments than the present study. Potential biological mechanisms include PM2.5-induced ACE2 overexpression and short-term cellular changes leading to increased susceptibility, though exposure duration requirements remain uncertain.

Design: Time-series analysis of daily SARS-CoV-2 test positivity at Renown Health (largest healthcare system in Washoe County, NV) from 15 May to 20 Oct 2020, encompassing a prolonged wildfire smoke event. Outcome: Daily count of positive SARS-CoV-2 nucleic acid amplification (NAA) tests at Renown with an offset for daily total tests to model positivity rate. Testing data: Included all NAA test types except one briefly used assay with very low positivity and minimal use during peak smoke; counted only the initial positive test per patient for positives; for totals, included first positive for positives and one negative per patient per day for those never positive (from 2 Mar to 21 Oct 2020). Exposure: Ambient PM2.5 from four EPA FEM BAM 1020 VSCC monitors in Reno/Sparks, aggregated to a daily weighted average with weights proportional to the count of Renown patients residing within 5 km of each monitor. Handling missing/invalid PM2.5: If a monitor was missing, used weighted average of remaining monitors; replaced six negative values with zero; recovered missing 29 Sep 2020 data from county AQMD. Meteorology: Temperature and relative humidity from KRNO airport station (5-minute data averaged to daily means). Missing meteorology handled by ignoring sporadic gaps; for 9 Aug 2020 (>25% missing), imputed as the average of adjacent days. Covariates: 7-day average of mean temperature (primary), day-of-week indicators, cubic regression spline of time (degrees of freedom constrained to 4 to avoid overfitting and negative residual autocorrelation), and an autoregressive term (lag-1 count of positives) to remove residual autocorrelation (validated by ACF/PACF and Durbin–Watson test). Statistical model: Generalized additive model with Negative Binomial distribution. Model form: log(Y_i) = β0 + β1 temp_i + β2 Y_{i-1} + β3 DOW_i + s(time_i) + β4 PM2.5_i + log(total_i) + ε_i, where log(total_i) is an offset, effectively modeling positivity rate. PM2.5 exposure windows: Initially examined lags 0, 7, 14 for daily mean PM2.5 and 7-day average PM2.5. Extended analyses evaluated lags 0–14 for single-day PM2.5, and lags 0, 3, 6, 9, 12 for 3-day averages. Distributed lag models (DLMs): Applied Schwartz’s polynomial DLMs, constraining lag effects to quadratic (lags 0–8) or cubic (lags 0–12), based on observed patterns (positive associations at intermediate lags, near-zero at longest lag). Excess case estimation: Using the 7-day average PM2.5 model (chosen a priori and parsimonious), simulated a counterfactual scenario for 16 Aug–10 Oct 2020 with PM2.5 set to the 2019 average for the same calendar period (4.5 µg/m³). Generated fitted values on the link scale (z_i), exponentiated and multiplied by daily test totals to obtain expected positives, and calculated excess cases as the difference from observed. Uncertainty quantified via resampling: for each day, sampled z_i from Normal with model-based SE, repeated 10,000 times to form empirical 95% CI. Sensitivity analyses: (1) increased time spline df max from 4 to 9; (2) added 7-day average relative humidity; (3) replaced 7-day average temperature with 3-day average. Software and oversight: R 3.6.0 with mgcv and mgcv.helper for modeling; ggmap for mapping. Study deemed IRB-exempt (UNR IRB #1106618-25).

- Study period: 159 days (15 May–20 Oct 2020); 35,955 individuals tested at Renown; 2,881 positives (8.0% positivity). - Age distribution: 72% were ≥30 years; highest positivity in ages 18–29 (11.3%). - Testing volume increased from ~130/day (late May) to ~404/day (early Oct). - PM2.5 exposure: 59 days affected by wildfire smoke; 50 days between 16 Aug–10 Oct. Daily mean PM2.5 (across monitors) ranged 1.5 to 114.3 µg/m³. - Primary association: Each 10 µg/m³ increase in the 7-day average PM2.5 was associated with a 6.3% relative increase in SARS-CoV-2 positivity rate (95% CI: 2.5% to 10.3%). - Lag structure: Positive associations for single-day PM2.5 at lags 2–6; for 3-day averages at lags 0 and 3. DLMs (quadratic lags 0–8; cubic lags 0–12) indicated significant positive effects at lags 2–6, approximately 1% relative increase in positivity per 10 µg/m³ per lag. - Excess burden: During 16 Aug–10 Oct 2020, wildfire PM2.5 exposure was estimated to increase cases by about 17.7% (95% CI: 14.4%–20.1%) at Renown, corresponding to roughly 178 additional positive cases. - Robustness: Sensitivity analyses (time spline df, adding humidity, alternative temperature averaging) yielded similar results.

The analysis indicates that elevated wildfire-related PM2.5 substantially increased SARS-CoV-2 test positivity in Reno during prolonged smoke episodes, supporting predictions that wildfire smoke can exacerbate COVID-19 transmission. Compared with prior studies during the 2020 wildfires, this study strengthens causal inference by adjusting for time-varying confounders (overall pandemic trend via a time spline), meteorology (temperature), testing volume (offset), and autocorrelation. The observed lag pattern (strongest associations 2–6 days prior) aligns with a short-term cumulative effect of PM2.5 exposure on infection risk, potentially via PM-induced immune modulation, enhanced viral deposition/transport, and/or ACE2 receptor upregulation. Non-biological mechanisms may also contribute: smoke events may alter human behavior (e.g., increased time indoors), possibly affecting transmission dynamics depending on public policy (open/closed indoor venues, school schedules). The findings are consistent with broader evidence linking air pollution to COVID-19 incidence and severity across multiple geographies. Overall, the results suggest that ambient PM2.5 surges from wildfire smoke can meaningfully amplify COVID-19 spread within days, underscoring the need for integrated wildfire smoke and pandemic response policies.

This study demonstrates a significant association between wildfire smoke PM2.5 and increased SARS-CoV-2 test positivity in Reno, NV, with an estimated 6.3% rise in positivity per 10 µg/m³ increase in 7-day average PM2.5 and approximately an 18% excess in cases during peak smoke exposure. By incorporating robust adjustments and lagged exposure analyses, the work substantiates air pollution’s role in exacerbating COVID-19 transmission. The findings can inform public health preparedness where wildfire smoke overlaps with respiratory pandemics, including lowering PM2.5 action thresholds during high SARS-CoV-2 prevalence, establishing socially distanced clean-air shelters, and ensuring access to appropriate respirators. Future research should parse pollutant-specific effects (beyond PM2.5), refine individual-level exposure assessment, evaluate morbidity and mortality attributable to smoke-related COVID-19 increases, and examine behavioral mediators and policy interventions.

- Observational design with potential residual confounding (e.g., unmeasured changes in human behavior, policy shifts). The time spline mitigates but does not eliminate confounding. - Exposure assessment based on outdoor monitor averages weighted by patient proximity; does not capture individual-level exposure variability due to occupation, indoor filtration, housing conditions, or time–activity patterns. - Focus on PM2.5 only; wildfire smoke contains other pollutants (e.g., coarse PM, CO, VOCs, ozone, NOx) that may confound or contribute to observed associations. - Did not assess clinical severity or mortality; only positivity rates at a single health system were analyzed. - Some potential influence of contemporaneous events (e.g., school reopening) cannot be fully disentangled, though temporal patterns suggest transient surges rather than sustained increases. - Model specification choices (e.g., degrees of freedom for time spline) may affect estimates; sensitivity analyses suggest robustness.