This prospective study aimed to identify preadmission factors — including demographic features and early symptom clusters — that were associated with COVID-19 outcomes. Outcomes examined included disease severity at recruitment, acute clinical complications during hospitalisation, survival, and long-term recovery. Analyses were performed both with and without controlling for an acute COVID-19 severity score at enrolment.
The research used a longitudinal, prospective, multicentre cohort design. Patients were recruited from four hospitals and one community general practice in England. Recruitment began in April 2020 at the start of the UK pandemic and continued through the first three waves, with the study completing in 2022.
A total of 425 patients with confirmed COVID-19 were enrolled. The cohort was skewed toward patients with severe disease at recruitment, with 60% classified as severe. Further demographic breakdowns beyond those reported below were not provided in the source.
Data collected were used to identify associations between preadmission demographic factors and symptoms and several outcomes: COVID-19 severity at recruitment, acute clinical complications (for example thrombosis, renal disease, pneumothorax), survival to discharge and hazard of death over time, and long-term recovery including persistent physical, cognitive and mood symptoms up to 12 months.
The authors applied multiple statistical approaches: logistic regression models, time-to-event (survival) analyses, mixed-effect models, and K-medoids clustering to define symptom clusters. Where reported, odds ratios (OR), hazard ratios (HR) with 95% confidence intervals (CI) and p values are presented; when source details were incomplete, that is noted.
A defined preadmission symptom cluster was associated with higher odds of specific acute complications. Patients in this cluster had higher odds of thrombosis (OR 2.7, 95% CI 1.6 to 6.1, p=0.021) and renal disease (OR 3.3, 95% CI 1.4 to 8.0, p=0.008).
Duration of symptoms prior to admission under 1 week was associated with increased odds of pneumothorax (OR 3.1, 95% CI 1.2 to 8.3, p=0.025). Additionally, renal complications were more frequent in the first pandemic wave than in the second (OR 3.4, 95% CI 1.5 to 7.5, p=0.003).
Using logistic regression to predict survival to discharge, the analysis reported that white (vs unknown) ethnicity was associated with greater odds of survival (OR 6.6, 95% CI 2.8 to 15.2). The absence of comorbidities was also associated with higher odds of survival (OR 2.7, 95% CI 1.2 to 5.9, p=0.016). The source reports an association involving age (OR 4.9, 95% CI 1.2 to 20.0, p=0.027) but the comparison or category for that odds ratio is not fully specified in the source text.
When survival was modelled using time-to-event (hazard) analysis, different relationships emerged. Short symptom duration before admission (less than 1 week) was associated with an increased hazard of death (HR 5.6, 95% CI 1.3 to 24.4, p=0.023). The presence of comorbidities was associated with a markedly increased hazard of death (HR 33.3, 95% CI 3.1 to 333.3, p=0.0035).
More than 80% of patients reported long-term sequelae. Early after illness onset, neuromuscular and cognitive effects were predominant. Up to 12 months after illness, mood disturbance and breathlessness persisted in many patients.
Gender was reported to have strong associations with multiple persistent symptoms. Associations included dyspnoea (OR 99.8, 95% CI 2.4 to 4123.1, p=0.02), persistent mood disorder (OR 8.3, 95% CI 1.7 to 39.9, p=0.008), and gastrointestinal and cognitive complaints: loose bowels (OR 3.3, 95% CI 1.1 to 10.8, p=0.045), abdominal pain (OR 5.0, 95% CI 1.4 to 18.1, p=0.014), constipation (OR 3.8, 95% CI 1.6 to 9.2, p=0.003), and cognitive problems (OR 2.3, 95% CI 1.1 to 4.8, p=0.024).
In this cohort, specific demographic factors and early infection symptom patterns were associated with acute complications, survival outcomes and persistent post-COVID sequelae even after accounting for disease severity at enrolment. The authors indicate that understanding the underlying disease mechanisms that explain these associations is necessary to guide targeted therapeutic strategies to prevent or manage these complications.
All findings and numerical estimates above are reported as presented in the source. Where the source text omitted comparator details for a reported odds ratio (for example the age-related OR), those comparator categories were not specified and are therefore not inferred here. The cohort was skewed to severe cases (60%), which should be considered when interpreting generalisability.