Figure 2 | Scientific Reports

Figure 2

From: Predictive usefulness of RT-PCR testing in different patterns of Covid-19 symptomatology: analysis of a French cohort of 12,810 outpatients

Figure 2

Effects of patient characteristics on RT-PCR results. We apply a propensity-scoring weight to each patient, to remove the testing-propensity confound. We report weighted counts/proportions/odds-ratios. (a) Representation of each characteristic in the whole cohort (blue diamonds), the RT-PCR+ cohort (black dot) and the RT-PCR− cohort (white dot). (b) RT-PCR results as a function of different characteristics in the tested population (cf Fig. 1c for categories). Odds ratios of being RT-PCR+ when belonging to a given subgroup, and p-values that these ratios are significantly different from 1 (using a Wald test). Patients without comorbidities nor smoking are more likely to be RT-PCR+, so do patients with anosmia, ageusia, anorexia, fever, fatigue, cough, myalgia. On the opposite, comorbidities, in particular respiratory diseases, and symptoms such as breathlessness and cardiopulmonary symptoms are associated with RT-PCR. (c) Same analysis as (b), within the subpopulation that has not been hospitalized. Results are similar. Table S1 provides all numerical data.

Back to article page