Background

Conjunctivitis is probably the most common [1]ocular manifestation of COVID syndrome, and specific signs and symptoms have been frequently reported [2,3,4]. However, it is still unclear whether ocular involvement is associated with viral shedding in the conjunctiva/tears or it is rather a secondary involvement induced by the systemic infection. In this study, we investigated the prevalence of patient-reported ocular symptoms, physician-detected ocular signs, together with anosmia/ageusia and conjunctival viral shedding in a cohort of patients affected by COVID-19 infection.

Materials and methods

We evaluated 53 patients hospitalized between February 25th and September 16th, 2020 at the San Raffaele Hospital, in Milan, Italy with a confirmed diagnosis of COVID-19 infection. Presence of interstitial pneumonia was confirmed in all 53 patients with computed tomography scan imaging [5]. This study did not include critically ill (i.e. intubated) patients. Paired (within 3 days) nasopharyngeal and conjunctival swabs were performed during hospitalization. The conjunctival samples were collected from both eyes using a single swab, which was then stored in a vial and analysed within 24 h. Swab specimens were processed by Cobas® SARS-CoV-2 Test (Roche), which detects ORF-1a/b and E gene regions on SARS-CoV-2 genome, designed to be used on the automated Cobas® 6800 Systems [6]. We also reported the cycle threshold (Ct) values, when available, which are a useful proxy of viral load [7].

A 4–item questionnaire was administered to patients in order to investigate symptoms of conjunctivitis (red eyes, sticky eyes, tearing, burning) and presence of ageusia or anosmia. Finally, patients were evaluated by a study physician for the presence of the following ocular signs: conjunctival hyperaemia, secretion, chemosis and epiphora; previous history of ocular diseases was also collected.

The study was conducted in accordance with the Declaration of Helsinki [8] and the evaluated patients are part of the COVID-19 institutional clinical-biological cohort (Covid-BioB registered on the ClinicalTrials.gov website: NCT04318366), whose study protocol was approved by the Hospital Ethics Committee (protocol number 34/int/2020). Informed consent was obtained according to the Ethic Committee guidelines. Patients’ demographic and clinical characteristics, treatment, need/mode of oxygen support as well as values of laboratory parameters during hospitalization were extracted from the Covid-BioB database.

Statistical analysis

Median values and quartiles (IQR), were used to describe continuous variables while frequencies and percentages were used for categorical variables. Prevalence of patients with ocular symptoms or signs or ageusia or anosmia in the overall sample were estimated with the corresponding 95% confidence intervals (95% CIs) using the modified Wald method [9]. Characteristics of patients with or without symptoms or signs were compared using the Chi-square or Fisher’s exact test for categorical variables, and the Wilcoxon rank sum test for continuous variables.

Multivariate general linear regression models were fit to estimate mean differences in continuous outcomes (laboratory parameters at conjunctival swab) comparing patients with vs without ocular symptoms or signs, adjusted for two potential confounders [use of steroids before or at ocular swab (yes vs no) and timing of ocular swab execution since hospitalization (continuous variable in days)].

Two-tailed P values are reported and a P < 0.05 considered to indicate statistical significance.

Statistical analyses were performed with the SAS Software, release 9.4 (SAS Institute, Cary, NC).

Results

Forty-six out of 53 patients showed a positive nasopharyngeal swab for SARS-CoV-2 infection at the time of conjunctival evaluation and were considered for further analysis. The median age was 65.0 years (48.1–78.0) and 24 (52%) were males. Four patients had a history of previous ocular diseases: 2 patients reported dry eye disease, 1 glaucoma and 1 ectropion.

All the conjunctival swabs were negative. Ocular symptoms were present in 17 [37%, 95% confidence interval (95%CI): 24.5–51.4] patients. Red eye was reported by 3 (7%) patients, sticky eyes by 4 (10%), tearing by 6 (15%), burning by 7 (16%). Patients’ characteristics according to the presence of ocular symptoms are reported in Table 1.

Table 1 Patients’ characteristics according to the presence of ocular symptoms
Full size table

The study physician observed ocular signs in 13 [28%, 95% confidence interval (95%CI): 17.2–42.7] patients: epiphora in 9 (20%) patients, secretion in 6 (13%), conjunctival hyperaemia in 3 (7%), chemosis in 0%; anosmia was reported in 11 (24%, 95% confidence interval (95%CI): 13.8–38.1] patients and ageusia in 13 [28%, 95% confidence interval (95%CI): 17.2–42.7].

Patients with ocular symptoms or signs tended to be older: 76.8 years (62.4–83.3) vs 57.2 years (48.1–74.0), p = 0.062 and had a longer hospitalization: 38 days (18–49) vs. 14 days (11–21), p = 0.005 (Table 1).

Importantly, at swab sampling, Ct values were similar between patients with signs or symptoms than those without them [29.5 (24.3–32.8) vs 32.5 (25.1–35.0), p = 0.432].

Plasma concentrations of Interleukin-6 were significantly increased in patients with signs or symptoms in comparison with those without them: 43.5 pg/ml (19.7–49.4) vs. 8 pg/ml (3.6–20.7), p = 0.02. Red cell distribution width was also significantly higher: 15 (14.3–16.7) vs 13.2 (12.4–14.4), p = 0.001. Finally, oxygen saturation, PaO2/FiO2 ratio or the number of patients treated with nasal cannula or Venturi mask or non-invasive mechanical ventilation were not affected by the presence of ocular signs or symptoms. Detailed information about patients presenting ocular signs or symptoms is reported in Table 2.

Table 2 Patients presenting ocular signs or symptoms
Full size table

At multivariate analysis, plasma concentrations of Interleukin-6 and red cell distribution width were on average higher in patients with signs or symptoms (IL-6 β: + 32.6, 95%CI: 1.33–64.0, p = 0.042; RDW β: + 2.15, 95%CI: 0.91–3.39, p = 0.001) compared to patients without them after adjusting for confounders; a marginally significant difference was seen in Ct values suggesting lower values (i.e. higher viral load) for patients with signs or symptoms when adjusting for confounders (Ct β: -4.32, 95% CI: − 9.30, 0.66, p = 0.086).

Discussion

In this paper, we show that conjunctivitis symptoms/signs are associated with COVID-19 disease course. This is supported by a number of observations. First, the finding of elevated levels of IL-6, an important predictor of disease severity [10] [11]. Second, patients with ocular manifestations showed lower Ct values (i.e. higher viral load) in rhinopharyngeal swabs. Third, we also found that patients with ocular symptoms had a longer hospitalization, suggesting a relationship between ocular symptoms and severity of the disease. However, no significant difference in patients’ age -one of the most important prognostic factor- was found between the two groups. Similarly, no significant difference was detected in the type of therapies administered during the hospital stay, which could have had an impact on viral clearance. Importantly, ocular signs/symptoms were not induced by ventilatory support, as our analysis failed to detect any association between the two.

Therefore, our study suggests that testing for ocular symptoms/signs could be a helpful tool to alert clinicians on a more severe clinical phenotype. In this vein, our results are corroborated by findings from Ping Wu et al. [12]

We did not find expression of COVID-19 RNA on the conjunctiva of our patients, similarly to others, who repeatedly tested a cohort of COVID-19 patients [13]. Azzolini et al [14] recently found virus expression in the conjunctiva, although clinical and methodological differences make comparisons difficult. However, the role of the ocular mucosa as a potential entry site or as a potential reservoir for the virus remains unclear.

We acknowledge some limitations including the cross-sectional design of the study, the limited number of patients and the timing of swab collection that varied as a consequence of the critical situation induced by COVID-19 in our hospital.

Importantly, ocular signs/symptoms were not induced by ventilatory support, as our statistical analysis failed to detect any association between the two.

Conclusion

In summary, we found that over a third of the patients had ocular signs or symptoms and about a quarter presented anosmia and ageusia. While additional studies are needed to confirm our findings, we propose that testing ocular signs and symptoms at hospitalization can be an effective, non-invasive and rapid screening measure of COVID-19 patients.