Geospatial Point-of-Care Testing Strategies for COVID-19 Resilience in Resource-Poor Settings: Rural Cambodia Field Study

Background Point-of-care testing (POCT) generates intrinsically fast, inherently spatial, and immediately actionable results. Lessons learned in rural Cambodia and California create a framework for planning and mobilizing POCT with telehealth interventions. Timely diagnosis can help communities assess the spread of highly infectious diseases, mitigate outbreaks, and manage risks. Objective The aims of this study were to identify the need for POCT in Cambodian border provinces during peak COVID-19 outbreaks and to quantify geospatial gaps in access to diagnostics during community lockdowns. Methods Data sources comprised focus groups, interactive learners, webinar participants, online contacts, academic experts, public health experts, and officials who determined diagnostic needs and priorities in rural Cambodia during peak COVID-19 outbreaks. We analyzed geographic distances and transit times to testing in border provinces and assessed a high-risk province, Banteay Meanchey, where people crossed borders daily leading to disease spread. We strategized access to rapid antigen testing and molecular diagnostics in the aforementioned province and applied mobile-testing experience among the impacted population. Results COVID-19 outbreaks were difficult to manage in rural and isolated areas where diagnostics were insufficient to meet needs. The median transit time from border provinces (n=17) to testing sites was 73 (range 1-494) minutes, and in the high-risk Banteay Meanchey Province (n=9 districts), this transit time was 90 (range 10-150) minutes. Within border provinces, maximum versus minimum distances and access times for testing differed significantly (P<.001). Pareto plots revealed geospatial gaps in access to testing for people who are not centrally located. At the time of epidemic peaks in Southeast Asia, mathematical analyses showed that only one available rapid antigen test met the World Health Organization requirement of sensitivity >80%. We observed that in rural Solano and Yolo counties, California, vending machines and public libraries dispensing free COVID-19 test kits 24-7 improved public access to diagnostics. Mobile-testing vans equipped with COVID-19 antigen, reverse transcription polymerase chain reaction, and multiplex influenza A/B testing proved useful for differential diagnosis, public awareness, travel certifications, and telehealth treatment. Conclusions Rural diagnostic portals implemented in California demonstrated a feasible public health strategy for Cambodia. Automated dispensers and mobile POCT can respond to COVID-19 case surges and enhance preparedness. Point-of-need planning can enhance resilience and assure spatial justice. Public health assets should include higher-quality, lower-cost, readily accessible, and user-friendly POCT, such as self-testing for diagnosis, home molecular tests, distributed border detection for surveillance, and mobile diagnostics vans for quick telehealth treatment. High-risk settings will benefit from the synthesis of geospatially optimized POCT, automated 24-7 test access, and timely diagnosis of asymptomatic and symptomatic patients at points of need now, during new outbreaks, and in future pandemics.


EIGHT FOCUS GROUP TOPICS FOR FEEDBACK ON COVID-19 POINT-OF-CARE (POC) STRATEGIES
Focus Topic 1-Cambodian Provinces in Highest Danger.Please identify the top five provinces or regions (e.g., international borders) in Cambodia (outside of Phnom Penh) with the highest risk of collapse of the local or regional healthcare systems from continuing and worsening COVID-19 contagion and overloading of local healthcare resources.
Focus Topic 4-Placement of Testing.Each province is characterized by its "small-world network," that is, the type of healthcare system in the context of the local culture, communications among people and professionals, the road network, the emergency medical system, ambulance routes, locations of emergency rooms, hospital capacity, availability of ICU beds (if any), and other key factors, such as mountainous geography.
Where in rural healthcare small-world networks, do you recommend COVID-19 testingmolecular diagnostic ("PCR"), rapid antigen ("Ag"), or antibody ("Ab") testing -should be placed?Please be specific and include a map of the highest risk province you have identified earlier with testing locations marked on the map of the province.
Identify the map you marked up:___________________________________________ Mark your placement of testing types as: "PCR," "Ag," or "Ab."

RAPID ANTIGEN TEST PERFORMANCE EVALUATION [2021]
Theory.PV GM 2 graphs help compare tiered sensitivity and specificity, governmentauthorized tests, commercial claims, and clinically evaluated rapid antigen tests (RAgTs).Tiered sensitivity/specificity comprise: T1) 90%, 95%; T2) 95%, 97.5%; and T3) 100%, ≥99%, respectively.Refinements and improvements in sensitivity and specificity help customize tests for clinical objectives.PV GM 2 pattern recognition reveals the significance of uncertainty when also considering 95% confidence intervals.PV GM 2 is not intended for point comparisons.Instead, please compare the performance configurations ("fingerprints") of the PV GM 2 curves.S1 and Table S2 present detailed summaries of performance from clinical evaluations, metastudies, and manufacturer claims.These tables provide the data for mathematical analysis using the equations in Table S3.The blue curves bracket the highest and lowest performance demonstrated in the clinical evaluations listed in Table S1.The purple line represents the median clinical performance from those evaluations.The clinical median lies remarkably close to the black curve reflecting the median of metastudies.Both exceed the WHO threshold of 80% for sensitivity, but the WHO threshold is too low to reliably rule out COVID-19 as prevalence increases.Rapid antigen tests with poor sensitivity generate false negatives, increasingly so as prevalence increases.

Figure S1. Roche Diagnostics Rapid Antigen Test Performance [2021]
The median clinical performance of the Roche Diagnostics RAgT does not match the manufacturer claim (red curve).However, the assay is somewhat balanced in that it displays a relatively "flat top" of the PV GM 2 curve in the low range of prevalence up to 40%.The manufacturer claims of PPA and NPA equal to 95.5% and 99.2% exceed Tier 2 thresholds for sensitivity and specificity of 95% and 97.5%, respectfully.However, in real-world clinical evaluations the median clinical performance of the Roche Diagnostics RAgT exceeds the Tier 2 curve only when prevalence is low.Thus, this RAgT will generate false negative results as the prevalence increases.

Figure S2. PanBio Rapid Antigen Test Performance [2021]
Figure S2 presents analysis of the Panbio COVID-19 antigen test (Abbott) based on the clinical evaluations in Table S2.Fall-off in clinical performance as prevalence increases is steep with a median that breaks below Tier 2 at very low prevalence, then falls precipitously.Hence, false negatives will increase exponentially as prevalence increases.In clinical evaluations the vertical range between high and low performance is somewhat limited but also positioned lower than that in Figure S1.The manufacturer claim of performance (red curve) does not achieve Tier 2 performance above 44.2%prevalence.The median clinical performance of the Panbio COVID-19 antigen test does not achieve the WHO threshold for sensitivity of 80%.
Cambodia.Two point-of-care formats became available in Southeast Asia in 2021, the Roche Diagnostics SARS-CoV-2 Rapid Antigen Test and the Abbott Panbio COVID-19 Antigen Test, for which Table

-The Role of COVID-19 Rapid Antigen Tests.
Diagnostic results from rapid COVID-19 antigen tests are available in a few minutes.Please describe a) the purpose of these tests in provinces with the highest prevalence of COVID-19 infections, b) the provinces where rapid antigen testing has been implemented already, c) the types of geographical sites within those provinces where the tests are located, and d) the manufacturers of the antigen tests.

Focus Topic 6-Selection of Instruments and Test Clusters for Mobile COVID-19 Testing.
Focus Topic 5-Mobile COVID-19Testing.Please describe when, where, and how mobile POC testing in a van or other vehicle or mobile health unit could be used in Cambodia.Identify provinces, actual testing sites, on-site sample collection,

Photo of the inside of the California van- Focus Topic 7-Instrumentation and Future Designs. We
should support health providers when they select COVID-19 diagnostics for different settings, environmental conditions, and community prevalence.Testing should be fast, so results are known quickly for quarantine, contact tracing, and possibly, hospitalization.Please describe your design of the IDEAL COVID-19 TESTING DEVICE.Be creative!Draw a picture!State the device specifications and the assays the device will perform.

Table S2 . Panbio Rapid Antigen Test Clinical Evaluations [2021]
Notes: 1. Median refers to sensitivities. 2. The Panbio COVID-19 rapid antigen detection test does not have US Food and Drug Administration Emergency Use Authorization status as of June 13, 2021.

:
Ag, antigen; CI, 95% confidence interval; COVID-19, Coronavirus Disease-19; Ct, cycle threshold; EUA, Emergency Use Authorization; FDA, Food and Drug Administration (USA); N, sample size; NP, nasopharynge4al; NPA, negative percent agreement; NR, not reported; POC, point-of-care; PPA, positive percent agreement; RT-PCR, reverse transcription-polymerase chain reaction; and TAT, turnaround time.PV GM 2 was created for visual logistics comparisons of performance curves of diagnostic tests, not for point comparisons.If the denominators of derived equations become indeterminate, then revert to the fundamental definitions, Eqs.1-6.The use of the formula for accuracy [Eq.33] is not recommended, because of duplicity of values with complementary changes in sensitivity and specificity.Equation Set Version 27.0 • August 27, 2023