Baseline characteristics
In total, 153 patients met the criteria for inclusion. The median age of the study population was 68 years (range 17–100). Of the 153 patients, 58 (38%) were female. Median time from symptom onset to hospital admission was 6 days (range 0–32). Of the 153 patients, 53 (35%) required transfer to the intensive care unit (ICU) during hospitalization. In-house mortality was 11% (17/153 patients) at the time of analysis (1st May 2020). According to the study criteria, 107 patients (107/153; 70%) were defined as confirmed COVID-19 (group A: PCR+, CT±) and 46 patients (46/153; 30%) as suspected COVID-19 (group B: PCR-, CT+). Table 1 illustrates the distribution of CT findings between group A and group B according to the level of certainty for COVID-19 using the in-house CT-based COVID-19 classification.
Table 1: Level of certainty for COVID-19 based on CT findings using in-house radiology classification
Level of certainty for COVID-19
|
COVID-19 CT classification
|
Group A (n=107)
% (abs.no)
|
Group B
(n=46)
% (abs.no)
|
Low
|
No signs of COVID-19 CT features (category 0)
|
4 (4)
|
-
|
Low
|
Infiltration or consolidation not typical for COVID-19 (category 1)
|
-
|
4 (2)
|
Low
|
Early stage of COVID-19 infection possible (category 2)
|
17 (18)
|
30 (14)
|
High
|
Typical CT features compatible with early COVID-19 (category 3)
|
23 (25)
|
20 (9)
|
High
|
Typical CT features compatible with advanced COVID-19 (category 4)
|
56 (60)
|
46 (21)
|
SARS-CoV-2 serology: results
99 of the 153 (65%) patients (65%) were SARS-CoV-2 seropositive. Of the seropositive patients, 77% (76/99) showed both IgM and IgG positivity, while 23% (23/99) of patients were only IgG positive. IgM and IgG seroconversions occurred in median 14 days (range 4–32) and 13 days (range 2–32) following SO, respectively.
Figure 1 displays the difference in seroconversion rates of SARS-CoV-2 serology distinctly for confirmed (group A) and suspected COVID-19 cases (group B). On days 5, 10, 15, 20, and 25 following SO, seroconversion rates of group A and group B were 8%, 25%, 65%, 76%, 91%, and 0%, 10%, 19%, 37% and 46% (p<0.01), respectively.
Figure 1: Illustration of seroconversion rates following SO in patients with confirmed COVID-19 (group A, solid blue line) and patients with suspected COVID-19 (group B, solid red line); p-value <0.01
In terms of severity of COVID-19 (Figure 2), seroconversion was higher and occurred earlier in non-ICU patients than in ICU patients. At 5, 10, 15, 20 and 25 days following SO, seroconversion was 8%, 22%, 64%, 74%, and 89%, respectively, in non-ICU patients, and 6%, 20%, 46%, 63%, 75%, respectively, in ICU patients (p=0.11).
Figure 2: Comparison of seroconversion rates over time between ICU patients (solid red line) and non-ICU patients (solid blue line); p=0.11
The impact of CT findings in relation to likelihood of COVID-19 based on seroconversion is demonstrated in Figure 3. Seroconversion rates of PCR-positive patients (group A) with CT findings compatible with a high level of certainty for COVID-19 according to the in-house CT classification were 9%, 26%, 65%, 77%, and 92% after 5, 10, 15, 20 and 25 days, respectively, following SO. In contrast, seroconversion rates of PCR-negative patients (group B) with CT findings consistent with high level of certainty for COVID-19 were 0%,10%, 20%, 40%, and 50% at the same time intervals following SO (p<0.01).
Figure 3: Difference in seroconversion between group A (solid red line) and group B (solid blue line); p <0.01. Analysis includes only patients with CT findings compatible with a high level of certainty for COVID-19
Seroconversion and PCR conversion were analyzed during hospitalization (Figure 4). In the early phase of infection, the detection efficiency of PCR for COVID-19 was higher compared with SARS-CoV-2 serology, whereas detection efficiency of serology for COVID-19 became superior >17 days following symptom onset.
Figure 4: Analysis of conversion rate of SARS-CoV-2 serology and SARS-COV-2 PCR test during hospitalization
In 12 of 46 (26%) suspected cases (group B), positive SARS-CoV-2 serology provided definitive diagnoses of COVID-19. All of these patients underwent at least two SARS-CoV-2 PCR tests (range 2–6). Eight of these 12 patients had typical CT features compatible with a high level of certainty for COVID-19 according to our COVID-19 radiological classification. Based on the findings by Wölfel et al.,6 which showed a COVID-19 seroconversion rate of 100% within 14 days after SO, patients from group A (confirmed COVID-19) and group B (suspected COVID-19) with at least one late serological test >14 days were separately analyzed in subgroup A1 and subgroup B1. Seroconversion rates of subgroup A1 (confirmed COVID-19) and subgroup B1 (suspected COVID-19) were 92% (59/64 patients) versus 52% (10/19 patients), resulting in a difference of 40%. In subgroup A1, seroconversion did not occur in 8% (5/64) of patients; all five patients suffered from immunosuppressive diseases such as active neoplasia (four patients) or diabetes mellitus (one patient). In comparison, 48% (9/19) of suspected cases with at least one SARS-CoV-2 serology test >14 days after SO (subgroup B1) had no seroconversion. Retrospective interdisciplinary clinico-radiolocial consensus re-evaluation of these nine patients revealed possible causes other than COVID-19 for their CT findings: one patient underwent bronchoalveolar lavage prior to CT scanning and developed lung edema due to iatrogenic fluid instillation; three patients showed signs of focal lung edema due to cardiac failure; one patient had pulmonary metastatic breast cancer with presumed lymphangiosis carcinomatosis, mimicking COVID-19; one patient`s CT showed signs of breathing artifacts; two patients had CT features in retrospect rather compatible with bacterial pneumonia (lobar pneumonia surrounding by ground-glass opacities with a parapneumonic pleural effusion) as a possible differential diagnosis. For the final one of these nine patients, no other rational differential diagnosis besides COVID-19 seemed to be reasonable.
Based on the results of the study, the CT-based triage algorithm for COVID-19 has been modified at our hospital during the pandemic, as illustrated in Figure 5.
Figure 5: Proposed modified CT-based triage algorithm at the University Hospital Klinikum rechts der Isar based on study results