During July 2–31, 2020, we administered a cross-sectional prevalence survey of patients attending 20 outpatient clinics, including hospitals and health centers, in 6 districts in Zambia (Appendix). The number of facilities we selected in each district was proportional to the number of facilities in the other districts (Appendix). We recruited participants from outpatient departments regardless of their reason for visiting the facilities. Study teams were instructed to recruit ≥50 participants per facility and to attempt to show no preference in selection. We obtained consent or assent (for minors) before beginning study procedures. Participants completed an interviewer-administered questionnaire that included demographics, medical history, SARS-CoV-2 exposures, history of recent illness, and reason for visiting the facility. Concurrently, we conducted a cluster-sampled household prevalence survey in the same 6 districts (5). These surveys provided an opportunity to directly compare outpatient and community SARS-CoV-2 prevalence estimates. The study was approved by the Zambia National Health Research Authority and the University of Zambia Biomedical Research Ethics Committee. The activity was reviewed by the US Centers for Disease Control and Prevention (CDC) and conducted consistent with applicable US federal law and CDC policy.

We tested nasopharyngeal specimens for SARS-CoV-2 RNA by using rPCR and plasma specimens for SARS-CoV-2 antibodies by using ELISA. We extracted RNA for rPCR using the QIAGEN Viral Mini procedure (https://www.qiagen.com). We used the Maccura SARS-CoV-2 Fluorescent PCR kit (https://www.maccura.com) on the QuantStudio 3 platform (ThermoFisher, https://www.thermofisher.com) as the primary rPCR diagnostic (6) and used the CDC assay method to confirm positive and indeterminant results (7). We performed the Euroimmun ELISA (PerkinElmer, https://www.perkinelmer.com) to test for spike protein IgG in single replicate (8). Participants could take part in any or all of the survey, rPCR testing, or serologic testing options based on personal preference.

We calculated SARS-CoV-2 prevalence as the number of positive results divided by the total number of tests conducted. Estimates were calculated separately for rPCR and ELISA results. We adjusted variance and 95% CIs for clustering by facility and seroprevalence for imperfect assay test characteristics (sensitivity 64.2%; specificity 100%; L. Steinhardt, pers. comm., email, 2021 Apr 2) using the Rogan-Gladen method (Appendix). To assess various factors associated with SARS-CoV-2 prevalence among outpatients, we used bivariate Poisson regression models to calculate prevalence ratios (PRs) and 95% CIs, with a random effects term for facility.

Of 1,975 persons representing ≈3.5% of ≈57,000 outpatients from the 20 facilities that we approached in July 2020 about participating (District Health Information System version 2; https://dhis2.org), 1,952 (98.8%) completed the questionnaire and 1,908 (97.7%) submitted either nasopharyngeal (1,490; 76.3%) or blood (1,657; 84.3%) specimens or both (Appendix). Of the 1,952 total participants, the number per district ranged from 160 (8.2%) in Nakonde District to 639 (32.8%) in Lusaka District; the median number of participants per facility was 93 (interquartile range 78–107; Table 1). Median participant age was 32 years (interquartile range 24–43 years); 60.5% were female. Overall, 34.1% of participants reported having a coexisting medical condition. Fever or respiratory complaints accounted for 28.2% of reasons for visiting the facility; 2.3% of participants were specifically seeking COVID-19 testing.

SARS-CoV-2 rPCR-measured prevalence was 13.4% (95% CI 8.3%–18.5%; Table 1); SARS-CoV-2 ELISA-measured prevalence was 8.2% (95% CI 5.1%–11.4%). Compared with community members, outpatients overall had higher rPCR-measured prevalence (PR 1.8, 95% CI 1.1–2.9; Table 2) as did those seeking COVID-19 testing (PR 3.6, 95% CI 2.2–5.9) or those without a stated reason for the visit (PR 2.0, 95% CI 1.2–3.3). Although only 2.2% of participants reported contact with confirmed COVID-19 case-patients, rPCR-measured prevalence was higher among outpatients specifically seeking COVID-19 testing compared with outpatients attending facilities for another reason (PR 2.2, 95% CI 1.4–3.3). In addition, outpatients had higher ELISA-measured prevalence than community members (PR 2.5, 95% CI 1.4–4.5) (Appendix). Among outpatients with SARS-CoV-2 infection, 45.7% did not report experiencing any symptoms associated with SARS-CoV-2.

Outpatients had higher SARS-CoV-2 prevalence than did community members in Zambia. Given the high SARS-CoV-2 prevalence and proportion of asymptomatic infections among outpatients, without instituting protective measures facilities might become transmission foci. Ameliorating risk requires instituting robust prevention and control strategies including universal masking in facilities (9,10). Furthermore, persons seeking testing at facilities should be quickly identified and isolated, because they might be at particularly high risk for having the virus.

One limitation of our study is that underlying condition and exposure history are subject to self-report and recall biases. The districts and facilities were not randomly selected and, despite our intentions to remain unbiased, may not have been representative of the population. The small sample size may have affected our ability to detect differences in factors associated with SARS-CoV-2 prevalence. The higher ELISA-measured prevalence among outpatients than community members could signal noncomparability between these 2 populations or that being an outpatient is a possible marker for other behaviors that increase SARS-CoV-2 infection risk. We assumed exact sensitivity and specificity values for the serology assay, but emerging evidence on serologic cross-reactivity (11–13) and antibody decay (14) might affect these values. However, given the timing of our study early in the outbreak, antibody decay was unlikely to substantially affect sensitivity (J. Perez-Saez, unpub. data, https://doi.org/10.1101/2021.03.16.21253710).

Countries with limited testing capacity need efficient strategies to identify persons with SARS-CoV-2 infections to interrupt transmission. In Zambia, when measured by rPCR, outpatients had 80% higher SARS-CoV-2 prevalence than persons in the surrounding community. Testing all outpatients regardless of their reasons for visiting the facility during periods of community transmission might help identify otherwise undetected SARS-CoV-2 infections. Compared with community-based SARS-CoV-2 testing, outpatient testing, which is often more convenient, might identify cases more effectively. Therefore, testing at facilities during periods of high community transmission might be an effective strategy to identify persons with SARS-CoV-2 infection, especially when testing capacity is limited.