3.1. Survey Results
The study sample included 99 respondents (93 P-ATS participants, 6 nonparticipants). Sample characteristics are shown in Table 2. Mean age was 20.36 years (SD 1.69). Table 3 provides an overview of self-reported anxiety levels within the total sample and across key groupings. Of respondents, 31.3% had high anxiety (score of >= 10) as measured by GAD-7 (43.3% of final-year and 26.1% of first-year students). Mean anxiety levels were higher in students living in private accommodation compared with students in university halls of residence. G*Power (version 3.1.9.7) [14] was used to calculate post hoc statistical power. All independent t-tests are underpowered (<0.8) and, therefore, the risk of false negatives was inflated and the results (including null findings) should be interpreted with caution.

3.1.1. Reasons for Testing
Students’ top three reasons for taking part were “helping to keep campus safe”, “contributing to the national effort to control the virus”, and “being involved in COVID-19 research” (see Figure 1). Table 4 provides an overview of participants’ experiences of the P-ATS. The majority of respondents reported they would take part in a COVID testing programme in the future (94.9% of P-ATS participants, 50% of non-participants), and would recommend university asymptomatic COVID testing to others (98% of participants, and 100% of non-participants). Reasons for nonparticipation were unrelated to COVID-19 or testing (e.g., not physically present at the university during this time) and there were no observable differences in demographics between participants and nonparticipants informed by descriptive statistics. Due to the small sample size group mean differences could not be tested using inferential statistics.

3.1.2. Case Identification
Only four positive SARS-CoV-2 cases were identified through the P-ATS in this sample. Three of these students reported that they were notified of their positive result within 24 h by the Clinical Virologist, and one student reported that they had been notified after 2 days. All four students were compliant in notifying the university the same day using an online reporting form. All received the official test kit and all self-isolated as advised. One of the students stayed exclusively in their own room during self-isolation, the other three self-isolated within their household but had contact with other household members.

3.1.3. Acceptability and Programme Satisfaction
Test kits were collected by individuals or members of their household and almost all of the participants did not report any issues with drop off and collection procedures. More than three quarters of respondents (79.6%) indicated they were confident in the outcome of their COVID-19 test result. Indicators of acceptability are shown in Table 4 and Table 5. In general, students were highly satisfied with the information they received about the testing programme (97.5%) and how the information was communicated to them (89.2%). Respondents were satisfied with the approach to communicating positive test results, but over one-third were dissatisfied with receiving negative test results via a group email (e.g., indicating that all individuals who tested positive had been informed), rather than being told their negative result individually (the process in place during the pilot deployment).

3.1.4. Testing Adherence
Adherence could be meaningfully determined for students in their first year who had participated in the P-ATS, as they were the target population and had all been offered the full P-ATS provision (testing provision for final year and staff was individualised, so highly variable). Reported adherence to testing related to COVID was relatively high among first year students. Of the first-year survey respondents who had participated in P-ATS (n = 65), 70.8% (n = 46) submitted all 10 swab tests in weeks 1–10 (full swab provision), and 93.9% (n = 61) submitted 5 or more swabs (at least half the swab provision). With regard to saliva samples only, 89.2% (n = 58) of first years completed one or more samples, and 16.9% (n = 11) completed two or more saliva samples. For both types of test combined, 92.3% (n = 60) completed 6 or more tests, and 47.7% (n = 31) completed all 12 tests (10× swab, 2× saliva). Reported PCR test completion is provided in Figure 2, Figure 3, Figure 4 and Figure 5 for the first year P-ATS participants (n = 65). The change from swab test to saliva sample was initiated at week 10 of 12, in preparation for alignment with deployment of the main university testing service. Engagement willingness may be underestimated from the number of tests completed, due to brief period of test kit stock depletion during the study period.
As would be expected, due to a longer period in the P-ATS, first-year students completed significantly more PCR tests than final-year students during the pilot: Xfirst year = 10.13, SD = 2.82, n= 65; Xfinal year= 2.68, SD = 1.12, n = 28; t (91) = 12.51, p < 0.001. Figure 4 stratifies these reported frequencies across year groups. Across the whole sample (n = 93), reported adherence to PCR testing was significantly higher in those who had been required to self-isolate at any point during the P-ATS (Xself-isolate= 9.03, SD= 2.98, n = 34; Xno self-isolate= 7.15, SD = 4.82, n = 59; t (91) = 2.19, p = 0.031, Xdifference: 1.99), and those with lower levels of anxiety (Xlow anxiety = 8.70, SD = 4.24, n = 64; Xhigh anxiety = 6.07, SD = 3.99, n = 29; t (91) = −2.83, p = 0.006). Adherence to PCR testing was also higher in those students who lived on campus (n = 63) compared with those who lived elsewhere (X = 10.15on campus, SD = 2.67, n = 61; Xoff campus = 3.56, SD = 2.99, n = 32; t(91) = −10.14, p < 0.001), although this is not unexpected, given that almost all of the on-campus students were in their first year (n = 62) and were offered the full 12 weeks of P-ATS. Among the first-year students specifically (n = 65), adherence to PCR testing was significantly higher in those with lower levels of anxiety (Xlow anxiety = 10.71, SD = 2.47, n = 48; Xhigh anxiety = 8.47, SD = 3.14, n = 17; t (63) = −2.98, p = 0.004).
With regards antibody tests, 76.9% (n = 50) of first-year students reported completing three or more of the six tests offered during the P-ATS, and 41.5% (n = 27) of first-year students completed all six tests. Figure 5 stratifies these reported frequencies across year groups.

3.1.5. Correlates of Self-Testing
Pearson and Kendall’s Tau-b correlation coefficients were calculated to examine the association between reported test compliance, mental health and programme satisfaction for all student participants in the P-ATS (combined PCR swab and saliva samples, Table 6; antibody tests; Table 7). Increased PCR test adherence was associated with increased worry about friends and family contracting COVID-19, greater satisfaction with drop-off location, increased satisfaction with how negative tests were communicated, greater satisfaction with the information received and greater satisfaction with how information was communicated. A higher frequency of completed antibody tests were associated with greater worry about friends and family contracting COVID-19, greater satisfaction with information received and how it was communicated, and satisfaction with drop-off location.
Due to the exploratory nature of this study, a stepwise entry method was used (with the exception of covariates) to specify regression analysis. The frequency of combined reported PCR testing (swab and saliva) completion was used as the dependent variable. Gender was dummy coded, with males set as the referent group. Covariates (gender and year of study) were entered in block one (forced entry), followed by student’s mental wellbeing variables (block two, stepwise entry), and satisfaction with programme services and communication (block three, stepwise entry). See Table 8 and Table 9 for overview. Two standardised residuals (>+/−3.0) were identified as extreme scores and were removed from this analysis (n = 91). Increased PCR test adherence was significantly associated with being in the first year of study (as expected due to the higher number of tests available for first years compared with final years), lower levels of anxiety, increased worry about contracting COVID-19 and increased satisfaction with the way in which information was communicated (see Table 8). The final regression model accounted for 78.2% of the explained variance in the dependent variable (adjusted R2 = 0.769, SE = 2.06). The statistical correlates associated with the frequency of antibody tests were determined with several study variables (see Table 7). Increased frequency of antibody tests completed during the pilot was statistically significantly associated with being first year of study (again, as expected due to the higher number of tests offered to first year compared with final years), lower level of anxiety symptoms, having greater worry about friends or family contracting COVID-19, increased satisfaction with the drop-off point for completed tests and greater confidence in the outcome of the test. The final regression model explained 57.9% of the total variance (adjusted R2= 0.579, SE = 1.54; see Table 9).

3.1.6. Correlates of Mental Health
A regression analysis was conducted to examine the correlates of mental health (specifically, anxiety measured by the GAD-7) and testing procedures or concerns regarding COVID-19. A three-step hierarchical regression was conducted using a stepwise entry method. Gender was dummy coded, with males set as the referent group. The final regression model accounted for 46.2% of the total explained variance: adjusted R2 = 0.431, SE = 4.48. See Table 10 for overview of regression results and block entry method. Increased anxiety among students was associated with (listed in descending order of magnitude of association strength): increased worry about contracting COVID-19, decreased satisfaction with the way in which information was communicated through the pilot and increased worry about friends and family contracting COVID-19.

3.1.7. Protective Health Behaviours
Perceived risk of COVID-19 (before and after the testing pilot) and the perceived importance of protective health behaviours is shown in Table 11 for the entire sample, and across sub-groups (positive tests, participants in pilot, and those that have self-isolated). Independent t-tests were calculated to test group mean difference. It is important to note that all t-tests were underpowered, with an elevated risk of false negatives (Type II error). Therefore, null results should be interpreted with caution. Those who opted not to take part in the pilot programme reported a lower perceived risk of COVID-19 than pilot participants pre- (July) and post-pilot (October) compared with those who had taken part, although the difference only reached statistical significance for the pre-pilot rating. There were nonsignificant trends towards lower perceived importance of protective behaviours in nonparticipants, those who had received a positive test result and those who had needed to self-isolate during the intervention period. However, it is important to note that, due to limited sample size of one comparison group, these tests have limited statistical power (resulting in an inflated risk of type II error) and should be interpreted with caution.