In March 2020, almost all United States (US) public schools closed their buildings in response to the COVID-19 pandemic and transitioned to virtual instruction to prevent the spread of COVID-19 and protect student, staff, and community health. Continued health surveillance and public health and education research provided evidence that students could safely return to in-person instruction with the implementation of COVID-19 prevention strategies [1–4]. Students returned to school buildings at varying time points during the 2020–2021 school year, with some schools remaining virtual the entire year and some schools opting for alternative schedules [5]. By the 2021–2022 school year almost all schools had returned to in-person learning [5,6], marking a unique school year in the timeline of the pandemic as there was still substantial transmission of SARS-CoV-2, the virus that causes COVID-19, and not all school-age children were eligible for vaccination [7]. Further, scientific evidence on the feasibility and effectiveness of individual strategies to prevent the spread of COVID-19 was still developing, and staff and parents were divided about strategies schools should implement [8,9].

When developing prevention plans, schools partnered with community organizations and local health and education departments, gathered feedback from staff and parents/guardians about the appropriateness and feasibility of key prevention strategies, and followed guidance released by local, state, and federal agencies. The Centers for Disease Control and Prevention (CDC) released and updated guidance that outlined infection prevention and control strategies schools could implement to reduce the spread of COVID-19 such as universal mask wearing, improving ventilation, social distancing, daily cleaning, and promoting vaccination [10]. This guidance encouraged schools to layer prevention strategies based on community context, stating that “localities should monitor community transmission, vaccination coverage, screening testing, and occurrence of outbreaks to guide decisions on the level of layered prevention strategies (e.g., physical distancing, screening testing)” [10]. Though several studies have examined implementation of and barriers to implementing specific strategies (e.g., mask wearing, physical distancing, testing) [11–19], to our knowledge there are no longitudinal studies that have examined implementation of recommended strategies during the 2021–2022 school year. The purpose of this study is to examine the implementation of strategies outlined in CDC’s guidance for schools during the 2021–2022 school year using a nationally representative sample of US K-12 public schools. This study also examines the extent to which school and community characteristics influenced strategy implementation.

To help account for nonresponse bias and produce estimates representative of K-12 public schools, we calculated weights for this analytic sample. We developed multivariable-adjusted logistic regression models to examine the independent effects of school characteristics associated with participation. School-level participation was modeled as a function of city, majority white, an affluence indicator (a socioeconomic status measure from MDR categorized as low or below average, average, above average or high) [22], census region, and school level (based on sampling strata). We developed nonresponse adjustment classes based on variables found to have a significant influence on school participation. The final weights were the result of post-stratification adjustments. For each post-stratum (jk), the weights can be expressed as the product of the post-stratification (PS), nonresponse adjustments (NR), and the school sampling weights. WFinaljk=Wjk*WNRjk*WPSjk

We calculated the weighted prevalence and 95 % confidence intervals (CIs) for each prevention strategy. We used generalized estimating equations (GEE) to allow for modeling of discrete outcomes with longitudinal data [27]. We ran separate weighted GEE models with each COVID-19 prevention strategy as the dependent variable and wave as the independent variable (Wave 2, the beginning of the school year, served as the reference). Models adjusted for school level, NCES locale, percent student body eligible for free and reduced-price meals, county-level SVI, and county-level COVID-19 positivity rate. Estimated coefficients were converted to adjusted odds ratios (aOR) for interpretability. To further investigate the trend throughout the school year, for each GEE model, we conducted a pairwise multiple comparisons test of the wave coefficients to test for significant differences in strategy implementation at all pairwise survey time points. These pairwise findings were adjusted using Tukey’s Honestly Significant Difference (HSD). Two-way interaction effects between survey wave and school level and NCES locale were individually considered to examine if changes in strategy implementation over time varied by these school characteristics. Quasi-likelihood under the independence model criterion (QIC) and correlation information criterion (CIC), criterion used for model-selection and correlation structure specification, provide evidence these interactions overfit the data and were therefore excluded from the final model [29,30]. Additionally, CIC was used to select the covariance structure of the data. An autoregressive (AR) order 1 was selected, indicating that strategy implementation was correlated over time within a school. Due to the number of statistical tests conducted and to control the false discovery rate, a Benjamini-Hochberg multiple comparisons adjustment was made to reported p-values [31]. All models were fit using R version 4.1.2 with the package geepack used for estimation [28].

Table 3 includes weighted prevalence and 95 % CIs for each prevention strategy by survey wave. The time periods for each wave are as follows: Wave 2 (October–November, 2021); Wave 3 (December 2021 – January 2022); Wave 4 (February–March, 2022); Wave 5 (April–May, 2022). Three strategies were implemented by a majority of schools (>50 %) across all waves—daily cleaning, opening doors/windows, and diagnostic testing. In Wave 2 and Wave 3, seven of the nine strategies were implemented by greater than 50 % of schools (i.e., mask requirement, opening doors/windows, daily cleaning, classroom distancing, diagnostic testing, contact tracing, quarantine). Fewer schools were implementing prevention strategies in Wave 4 and Wave 5; only opening doors and windows, daily cleaning, and diagnostic testing were implemented by greater than 50 % of schools in Wave 5.

Tables 4 and 5 show aORs, 95 % CIs, and adjusted p-values for the association between survey wave (time) and each prevention strategy. Classroom distancing, mask requirements, and quarantine significantly declined at each survey wave. Relative to Wave 2, the odds of classroom distancing were lower in Wave 3 (aOR=0.66; 95 % CI, 0.50–0.88), Wave 4 (aOR=0.42; 95 % CI, 0.31–0.56), and Wave 5 (aOR=0.20; 95 % CI, 0.15–0.28). Relative to Wave 2, the odds of a mask requirement for students and staff were lower in Wave 3 (aOR=0.64; 95 % CI, 0.50–0.82), Wave 4 (aOR=0.30; 95 % CI, 0.23–0.40), and Wave 5 (aOR=0.02; 95 % CI, 0.01–0.03). Relative to Wave 2, the odds of requiring students to quarantine were lower in Wave 3 (aOR=0.33; 95 % CI, 0.24–0.45), Wave 4 (aOR=0.16; 95 % CI, 0.11–0.22), and Wave 5 (aOR=0.09; 95 % CI, 0.06–0.13). Results from the pairwise multiple hypothesis tests confirm the odds of classroom distancing, mask requirements, and quarantine were significantly lower at each survey wave throughout the year (Table 6).

Conversely, the odds of on-campus vaccinations were higher in Wave 3 (aOR=1.84; 95 % CI, 1.39–2.45), Wave 4 (aOR=2.09; 95 % CI, 1.61–2.71), and Wave 5 (aOR=1.59; 95 % CI, 1.20–2.11), relative to Wave 2. Increased odds of on-campus vaccination in Waves 3–5 were only significant when compared to Wave 2. The odds of conducting screening testing were 2.01 (95 % CI, 1.36–2.97) times higher in Wave 4 compared to Wave 2. The implementation of contact tracing in schools varied throughout the school year, compared to the beginning. The odds of contact tracing in schools were 2.07 (95 % CI,1.48–2.90) times higher in Wave 3 compared to Wave 2. By the end of the school year (Wave 5), the odds of contact tracing were 0.39 (95 % CI, 0.28–0.55) times lower than the beginning.

Pooling data across waves, several prevention strategies varied by school and community characteristics. The odds of on-campus vaccinations were 2.61 (95 % CI, 1.56–4.38) times higher for high schools than elementary schools. Compared to cities, schools in towns were 0.24 (95 % CI, 0.11–0.53) times as likely to have a mask requirement and schools in rural areas were 0.20 (95 % CI, 0.10–0.40) times as likely to have a mask requirement. The odds of conducting screening testing were 0.38 (95 % CI, 0.19–0.77) times lower in schools in rural areas compared to schools in cities.

Implementation of COVID-19 prevention strategies evolved over the 2021–2022 school year. Few data sources exist for national estimates of COVID-19 prevention strategy implementation in US public schools, but results from this study align with the US Department of Education’s School Pulse Panel in several ways: 1) common prevention strategies included increased cleaning, classroom distancing, and improved ventilation (e.g., opening doors/windows); 2) implementation of diagnostic testing was more common than screening testing; and 3) implementation of mask requirements declined during the Spring of 2022 [32]. Results from this study also show several strategies were consistently implemented across the 2021–2022 school year (i.e., cleaning, opening doors and windows, diagnostic testing) while other strategies increased initially and then declined (i.e., contact tracing, screening testing, on-campus vaccination) or declined consistently throughout the school year (i.e., mask requirement, classroom distancing, quarantine). Strategies with consistent implementation were more frequently implemented, for example, daily cleaning and opening doors and windows were implemented by 75 % of schools at each wave. Given the higher levels of consistent implementation, these strategies may be particularly feasible for schools to address a range of infection and prevention control needs over time.

Guidance released by CDC emphasized the need for localities to monitor the community context to guide decisions about the level of layered prevention strategies, suggesting considerations such as community transmission, vaccination coverage, screening testing, and occurrence of COVID-19 outbreaks [10]. Some of the patterns noted in this study align with expectations based on nationwide trends in COVID-19 cases and available resources. With the surges of Delta and Omicron variants of SARS-CoV-2, there was a general increase in COVID-19 cases nationwide from the start of the school year (August/September 2021) through January of 2022, and findings demonstrate higher implementation of several strategies during the first half of the school year with an increase in implementation of screening testing, on-campus vaccination, and contact tracing. Increases in implementation of on-campus vaccination and screening testing might also be partially attributable to increases in supply, availability (i.e., after the approval of the COVID-19 vaccine for youth ages 5–11 on October 29, 2021) [33], and funding [34,35].

Even though several trends aligned with expectations based on transmission levels, several strategies declined steadily throughout the year, even during periods of high COVID-19 transmission. Schools had to account for many interdependent factors when developing and updating prevention plans including public health guidance recommendations and updates, community transmission levels (e.g., COVID-19 cases, COVID-19 test positivity), and feasibility of implementation impacted by considerations such as available resources (e.g., vaccine availability by age group) and state support. Any of these factors could have contributed to schools’ decisions to revise prevention plans. Retaining or reintroducing some of these strategies might have been advantageous to minimize the spread of COVID-19, but it is possible it was too challenging for schools to re-introduce strategies that had been scaled back. It is also possible that some strategies (e.g., mask requirements) became regulated by external policies (e.g., district, state policies) that prevented their implementation. Previous research has noted several factors that impact feasibility of prevention strategy implementation including community and parental support, school infrastructure such as funding and personnel for resource intensive strategies, and availability and accessibility of resources, equipment, or health supplies [36]. More research is needed to determine what supports or resources schools might need to modify prevention plans (i.e., when to implement certain strategies and how to layer strategies), especially during periods of high transmission of respiratory viruses.

Although there was an increase in available resources across the 2021–2022 school year, on-campus vaccination and screening testing were not implemented by the majority of schools at any point during the school year. It is possible schools were encouraging vaccination by partnering with community providers or promoting district-wide vaccination events as opposed to school-level events, as measured in the current study. Screening testing was the strategy implemented least often, peaking at only 18 % of schools in February–March of 2022. Research has documented several barriers to implementing screening testing programs including limited perceived advantages when weighted against the perceived burdens, challenges obtaining consent, and overburdening staff who already have a demanding set of responsibilities [37]. Schools that have implemented higher resource strategies such as screening testing, on-campus vaccination, and contact tracing note the importance of strong community partnerships (e.g., local public health) [36–39].

Additional studies have found similar disparities in implementation of COVID-19 prevention strategies in rural areas [40–48], with political ideology [45,46] and misinformation or mistrust of public health officials as factors that account for some of the variation [40]. Identifying trusted community members (e.g., faith leaders) to disseminate accurate, up-to-date scientific information is one strategy that can build on the assets of rural communities [45]. Along with locale differences, high schools were more likely to report on-campus vaccination, and a previous NSCPS study also found school level differences in vaccination practices such as tracking vaccination status and providing information on COVID-19 vaccines to students [23]. These findings might be the result of vaccinations being approved earlier for youth ages 12 +; parents being more supportive of older children getting vaccinated; and/or district-wide vaccination clinics potentially being held at high schools due to the available size and space. Community and parental support and school infrastructure (e.g., presence of school nurse or school-based health center) will continue to be critical components to consider when supporting schools during public health emergencies.

These findings are subject to limitations. First, the study assessed the presence of prevention strategies but not factors like adherence and fidelity. Second, the response rate for participants who completed all four survey waves during the 2021–2022 school year was low (21 %). While we did create survey weights to address survey nonresponse, this likely cannot account for the self-selection bias and social desirability associated with this data collection. It is possible schools that responded to all four survey waves were more likely to implement prevention measures, and thus these findings may not be representative of all US K-12 public schools. Third, the study did not account for district, state, or federal influences such as state policies (e.g., mask mandates), which might have affected school’s ability to implement specific strategies. Finally, the study did not continuously monitor strategy implementation, making it challenging to determine exactly when schools modified their layered approaches.

Despite these limitations, this study furthers our understanding of COVID-19 prevention strategy implementation in US public schools during the 2021–2022 school year, providing an account of strategy implementation across various stages of the pandemic (i.e., variant waves like Delta and Omicron, approval of COVID-19 vaccine for youth 5–11 years, declining COVID-19 cases). Opening doors and windows and daily cleaning were consistently implemented by at least 70 % of schools, suggesting that these may be among the most feasible of strategies for schools to implement. Continued research examining how to provide infection prevention and control support to schools that prioritizes scientific evidence and allows for flexibility based on community context and school infrastructure will better prepare schools and their communities for future public health emergencies.