Background: Despite the fact that the clinical efficacy of hydroxychloroquine is still controversial, it has been demonstrated in vitro to control SARS-CoV-2 multiplication on Vero E6 cells. In this study, we tested the possibility that some patients with prolonged virus excretion could be infected by less susceptible strains. Method: Using a high-content screening method, we screened 30 different selected isolates of SARS-CoV-2 from different patients who received azithromycin \pm hydroxychloroquine. We focused on patients with viral persistence, i.e., positive virus detection in a nasopharyngeal sample \ge10 days, and who were tested during two French epidemic waves, late winter-spring of 2020 and the summer of 2020. Dose-response curves in single-molecule assays with hydroxychloroquine were created for isolates with suspected reduced susceptibility. Genome clustering was performed for all isolates. Results: Of 30 tested strains, three were detected as replicating in the presence of azithromycin + hydroxychloroquine, each at 5 \muM. The dose-response model showed a decrease in susceptibility of these three strains to hydroxychloroquine. Whole genome sequencing revealed that these three strains are all from the second epidemic wave and two cluster with isolates from Africa. Conclusions: Reduced susceptibility to hydroxychloroquine was not associated with viral persistence in naso-pharyngeal samples. Rather, it was associated with occurring during the second epidemic wave, which began in the summer and with strains clustering with those with a common genotype in Africa, where hydroxychloroquine was the most widely used.