Dissecting clinical features of COVID-19 in a cohort of 21,312 acute care patients

Categories: Disease & Virus

April 26, 2025

Study cohort

We analyzed 21,312 patient records across 47,063 retrospective encounters from five hospitals in Central Texas between March 2020 and September 2022 with the demographic features of the cohort displayed in Supplementary Table 1 and Fig. 2. The 30-day mortality rate for all evaluated patients was 3.23%, with a mortality rate of 7.73% among only admitted patients, and a 17.96% mortality rate among only severe/critical patients (TG1B-TG4).

**Fig. 2: COVID-19 trajectory groups capture disease severity and reveal underlying demographic differences.**

We identified four initial TGs on the modified WHO ordinal scale¹² over 30-days (Fig. 2A). To better capture acute hospitalization trajectories, TG1 was further divided into TG1A, TG1B, and TG1C by repeating modeling over only 12 days (Supplementary Fig. 2). The TGs were reflective of patient time to discharge, peak ordinal score, and time to peak ordinal score (Fig. 2B–D). We used the following patient classification system – TG1A: patients with mild to moderate disease who were not admitted or were quickly discharged (median discharge of 0 days); TG1B: severe patients who required a moderate length of stay (median discharge of 7 days); TG1C: severe patients with a longer hospital stay (median discharge of 11 days); TG2: critical patients who discharged by day 30 (median discharge of 16 days); TG3: critical patients who were not discharged by day 30 (with a median discharge of 39 days); TG4: all patients died within 30 days. TG1 included 91.63% of all patients, with TG1A, TG1B, and TG1C comprising 82.00%, 6.34%, and 3.29%, respectively. TG2, TG3, and TG4 represented 3.35%, 1.78%, and 3.23% of the cohort.

TGs also displayed significant association with patient demographics including sex, ethnicity, race, and age (Fig. 2E, I, Supplementary Table 1). Of note, males had worse outcomes, representing the majority of TG3 and TG4, while females had the highest representation in mild TG’s. Interestingly, patients documented as Latino/Hispanic were over-represented in TG3, however this difference was not observed in other TGs. There was no significant difference in race between TGs (Supplementary Table 1, Fig. 2G). TG4 patients had the highest median age of 70, while patients in TG1A had the lowest median age of 42 (Fig. 2H). Notably, TG1B-TG3 had similar median ages ranging from 57-61.

When comparing patient-reported SARS-CoV-2 vaccinations, the mild TG1A had higher vaccinations compared to severe/critical TGs (Supplementary Table 1). The lower severity groups TG1B and TG1C also had higher vaccination rates compared to critical groups TG2-TG4 (Supplementary Table 1).

Trajectory groups across COVID-19 waves

Based on the total COVID-19 cases in Austin’s Travis County, we classified five different outbreaks (herein referred to as waves) between March 2020 to September 2022 (Fig. 3A), roughly corresponding to Alpha/Beta (waves 1–2), Delta (wave 3), and Omicron (wave 4–5) (Fig. 3A), with the total number of hospitalizations reflective of the waves (Fig. 3B).

**Fig. 3: Trajectory groups over time.**

We observed significantly higher prevalence of TG1A in waves 3, 4, and 5 compared to waves 1 and 2 (Fig. 3C, Supplementary Data 1). Additionally, despite higher total case counts during the Omicron spike (Wave 4), total severe and critical hospitalizations for TG1B-TG4 were comparable between all waves (Fig. 3D). Age-associated mortality also decreased across the waves (Fig. 3F).

In waves 4–5 (Omicron), we noted a reduction in pulmonary/critical care, palliative care, infectious disease, nephrology, and neurological consultations and a decrease in diagnoses of pneumonia, acute respiratory distress, type-2 diabetes, and hypertension, (Supplementary Data 2–3) supporting the trend of a higher proportion of mild patients (TG1A) in waves 4–5. Interestingly when evaluating only the severe/critical patients (TG1B-4), neurological and psychiatric consultations remained stable across all waves, suggesting that concern of neurological involvement in severe disease remained persistent across the different circulating SARS-CoV-2 strains (Fig. 4).

**Fig. 4: Neurological and psychiatric consultations overtime.**

Diagnoses, hospital consultations, imaging studies, and medications across trajectory groups

Diagnoses

The comparison between the fatal (TG4) and most critical TG (TG3) revealed that TG4 patients had higher rates of cardiovascular co-morbidities, such as atherosclerotic heart disease, cardiac arrest, atrial fibrillation, nicotine dependence, and long-term aspirin use (Supplementary Data 4). Additionally, chronic kidney disease, cardiac arrest and chronic obstructive pulmonary disease were also elevated, implicating multi-organ damage in fatal patients compared to critical patients. Compared to TG4, TG3 patients were more likely to be anemic and experienced more electrolyte abnormalities and mixed acid base disorders. TG3 were also more likely to have ventilator dependence, sepsis, unspecified anxiety disorder, and dysphagia.

For the less severe groups, compared to TG1B-TG1C, TG1A patients had the least co-morbid diagnoses (Supplementary Data 4). Interestingly, TG1A patients had the highest association with headache and cough. Notably, alcohol related disorders were significantly higher in TG1B and TG1C patients compared to TG1A. Further, rheumatoid arthritis (RA) and inflammatory bowel disease (IBD), were significantly associated with TG1B-TG1C compared to TG1A. Additionally, severe/critical groups (TG1B-TG4) collectively had higher rates of several neurological and psychiatric disorders, including Alzheimer’s disease, cerebral infarction, epilepsy, migraines, myoclonus, neurocognitive disorder with Lewy bodies, non-traumatic intracerebral hemorrhage, Parkinson’s disease, restless leg syndrome, adjustment disorders, substance use related disorders, delirium, delusional disorders, depressive episode, generalized anxiety disorder, panic disorder, post-traumatic stress disorder, schizoaffective disorder, and schizophrenia (Fig. 5A). However, no neurological or psychiatric disorders were significantly different between the most critical TGs (TG3 vs TG2 or TG4 vs TG3).

**Fig. 5: Diagnosis, consultation, and imaging differences between COVID-19 TGs.**

Consultations & imaging studies

Although all consultations increased with higher TGs, TG4 received less critical care, infectious disease, gastroenterology, neurology, physical medicine and rehabilitation, hematology-oncology, psychiatry, and urology consultations compared to TG3 (Supplementary Data 5). Notably, TG3 and TG4 patients had the highest rates of palliative care and nephrology consultations, although only 64.20% of TG4 received a palliative care consult compared to TG3 (71.76%).

Notably, TG3 had the most imaging studies (Supplementary Data 6). Overall, most patients received respiratory imaging with increasing orders across TGs. Abdominal X-rays were significantly increased in more severe groups TG1B/TG1C compared to TG1A and in TG3 versus TG4. Furthermore, lower extremity venous ultrasound was also more common with increasing TG severity. Interestingly, 10.7% of the entire cohort received a Head CT without contrast, with frequency increasing from TG1A (7.6%) to TG3 (38.7%) (Fig. 5, Supplementary Data 6).

Medication administration

The least medication administration was in TG1A, likely due to fastest discharge (Supplementary Data 7). Notably, dexamethasone usage was not different between TG3/TG4, while Remdesivir was higher in TG3. Higher medication usage in TG4 compared to TG3 included haloperidol, morphine, lorazepam, scopolamine, glycopyrrolate, which are often administered with comfort-care measures, in line with higher palliative care consultations and’do-not-resuscitate’ orders in TG4. Additionally, neuromuscular blockers, atracurium, rocuronium, and succinylcholine as well as fentanyl, etomidate, and propofol were used more often in TG3 than TG4, correlating with a higher proportion of mechanical ventilation.

Laboratory tests and vital signs

Several laboratory tests demonstrated significant separation of TGs at admission and through hospitalization (Fig. 5B–G). Interestingly, albumin was one of the best predictors of disease severity, displaying the most ordinal relationship across TGs, both at admission and throughout hospitalization (Fig. 4C, Supplementary Data 8), with lower values in the more severe TGs, in correspondence with higher malnutrition diagnosis in the critical TG3. On admission (day 0-1), several other predictive labs distinguished the mild TG1A group from the more severe groups. TG1A had higher red blood cell count, hematocrit, and hemoglobin compared with more severe TGs (Supplementary Fig. 3). In addition, TG1A had elevated absolute basophils, eosinophils, lymphocytes, lower absolute neutrophils and immature granulocytes compared to more severe TGs and lower D-Dimer, sedimentation rate, lactate, C reactive protein and fibrinogen as well as higher calcium compared to the more severe TGs (Supplementary Data 8 and 10). Furthermore, in comparing the critical and fatal TGs, TG4 had elevated AST, total bilirubin, creatinine, BUN and anion gap (AGAP), and lower total protein and CO2. Finally, the fatal TG4 had higher lactate, AGAP and lower arterial pH compared to critical TG3.

Vitals signs from the first two days were averaged for each patient to identify changes across TG (Supplementary Data 9 and 11). Notably, systolic and diastolic blood pressure and oxygen saturation decreased with severity along with higher respiratory rate and lower Glasgow coma scores (GCS). However, only GCS was significantly different between TG4 and TG3. Over hospitalization, TG3/TG4 continued to have abnormalities in oxygen saturation, respiratory rate, and GCS to day 30 (Supplementary Fig. 4).

Source link