Factors associated with readmission to the Emergency Department in a cohort of COVID-19 hospitalized patients

2.1 Design and setting

This is an observational longitudinal retrospective study, conducted in a cohort of 441 patients who required hospitalization due to COVID-19 in the San Cecilio University Hospital (Granada, Spain) during the first wave of the pandemic, from 1 March to 15 April 2020. According to the inclusion criteria, only patients with confirmed positive polymerase chain reaction (PCR) to SARS-CoV-2 were considered in the study. The sample was followed up for 6 months from the date of discharge, thus follow-up finished on 10 January 2021.

2.2 Data source and variables

Information was collected from medical records of hospitalization and follow-up reports from Primary Care institutions during the 6 months following hospital discharge. All data were gathered by the same researcher (AR-D) to avoid heterogeneity of information and reviewed by two experienced researchers (MR-I and AC-C). The variables considered for the analyses were:

• Sociodemographic variables: sex, age, residence, dependence in activities of daily living.

• Clinical variables: comorbidities, laboratory workup, prognostic scores (CURB-65), treatment received, in-hospital mortality.

• Post-discharge symptomatology: information on symptoms during the 6 months after discharge were collected by means of periodic telephone consultations in Primary Care centers. These symptoms were classified into the following categories:

$\bullet$ Systemic symptoms: fever, asthenia, muscle weakness, musculoskeletal pain, general malaise.

$\bullet$ Respiratory symptoms: dyspnea, rib pain, thoracic pain, persistent cough, pharyngeal symptoms.

$\bullet$ Neurological symptoms: polyneuropathy, headache, sensitivity disorders, movement disorders, confusion, persistent anosmia or dysgeusia.

$\bullet$ Mental health symptoms: depressive, anxiety or sleep disorders.

$\bullet$ Hematological symptoms: anemia, thrombotic disorders.

$\bullet$ Dermatological symptoms: eczema, exanthema, pruritus, alopecia.

$\bullet$ Nephrological symptoms: renal failure.

$\bullet$ Urological symptoms: voiding syndrome.

$\bullet$ Ophthalmologic symptoms.

$\bullet$ Digestive symptoms: nausea, vomiting, diarrhea, constipation, anorexia, abdominal pain.

$\bullet$ Syncope or hypotension.

$\bullet$ Infectious symptoms: superinfections.

• Outcome: readmission to the ED within 6 months after hospital discharge.

2.3 Statistical analyses

First, a univariate descriptive analysis was performed. Then, bivariate analyses were applied using Pearson’s chi-squared and Student’s T-tests if the assumptions of parametric testing were met (otherwise, non-parametric tests were applied, e.g., Fisher’s exact test). The dependent variable of our study was the primary outcome, i.e., readmission to the ED during 6 months after discharge. Finally, multivariate analyses for predicting the outcome were performed using logistic regression models. The estimates of strength of association were calculated as odds ratios (ORs). All analyses were performed using the SPSS ® version 22.0 statistical software.

2.4 Ethical considerations

This study complies with the principles of the Declaration of Helsinki. A completely anonymized database was used for the analyses. No identification data were used. The study was approved by the Provincial Research Ethics Committee of Granada on 1 October 2020 (code 1585-N-20).

3.1 Descriptive analysis

Of the total sample of hospitalized patients (n = 441; mean age, 66.4 years; 55.1% men), 361 (81.9%) survived and were discharged. This group comprised the sample followed (mean age, 62.4 years; 52.9% men). A total of 75 (20.8%) patients were readmitted to the ED within 6 months after hospital discharge. The main characteristics of the subgroups of the cohort are described in Fig. 1. Table 1 shows the distribution of the sociodemographic and clinical variables of the followed sample, where the ‘total followed-up cohort’ column presents all the patients who were discharged alive, ‘readmission to the ED’ explores the subgroup of patients who were readmitted to the ED within the 6-month follow-up after discharge, and ‘no readmission to the ED’ describes the subgroup of patients who did not return to the ED during the said follow-up.

Fig. 1.Main characteristics (percentage, sex and age) of the subgroups of the cohort.

3.2 Bivariate analyses

As shown in Table 1, no significant differences in most of the sociodemographic or clinical variables at baseline were associated with readmission to the ED at 6 months, although a higher presence of comorbidities in the cohort of patients readmitted to the ED is observed for most of the analyzed variables. Conversely, not living at home and polymedication were significantly associated with the primary outcome. Table 2 shows the presence of symptoms during the 6-month follow-up after discharge of survivors (n = 361), stratified by the primary outcome. The ‘readmission to the ED’ column explores the subgroup of patients who returned to the ED during follow-up, and ‘no readmission to the ED’ describes the subgroup of patients who were not readmitted to the ED during follow-up.

As shown in the table, the most frequent symptoms were respiratory, systemic, neurological, and digestive.

Bivariate analyses showed that the factors potentially associated with readmission to the ED were fever, general malaise, thoracic pain, headache, hematological and nephrological symptoms, syncope or hypotension and superinfection. On the other hand, anosmia or dysgeusia and asthenia were associated with a lower frequency of readmission to the ED at 6 months.

3.3 Multivariate analysis

Multivariate logistic regression models were applied for predicting readmission to the ED (dependent variable). Models were adjusted for sex, age and the variables resulting from the bivariate analyses (Table 3). The ‘cOR’ column shows the crude associations between each variable and the odds of returning to the ED during follow-up, and the ‘aOR’ column shows the adjusted associations in the multivariate models.

After adjustment of the models, the variables that remained as risk factors for readmission to the ED at 6 months following hospital discharge were polymedication, not living at home, general malaise and thoracic pain. Fig. 2 shows the distribution of the main factors associated with readmission to the ED in the multivariate analysis.

Fig. 2.Main factors associated with readmission to the Emergency Department in the multivariate analysis. Percentage distribution and P-values.

4.1 Symptomatology

The frequency of persistent symptoms during 6 months after hospital discharge in our study was surprisingly high. Almost two thirds of the patients (64.5%) showed at least one symptom at follow-up. The most frequent symptoms identified in this study were respiratory (63.0%), especially dyspnea (44.9%); systemic (54.9%), especially asthenia (36.1%); neurological (36.7%); digestive (26.6%), especially diarrhea (17.7%); and related to mental health (22.0%).

The high frequency of potentially severe symptoms such as superinfection (8.0%), hypotension (4.7%) and hematological (3.3%) symptoms should also be considered.

In addition, the high frequency (10.5%) of dermatological symptoms associated with COVID-19 is in agreement with other studies [21, 22].

4.2 Factors associated with readmission to the Emergency Department

Readmission to the ED is an important concern which, considering the state-of-the-art literature on COVID-19, merits further research due to a number of reasons. First, during the pandemic, EDs have been overloaded due to an excessive healthcare demand. Second, readmission to the ED during the 6 months after hospital discharge might be a marker of severity in the short-to-medium term and possibly requires an optimization of the criteria used for hospital discharge or follow-up protocols of these patients by Primary Care physicians. We believe that the identification of factors associated with this severity might be useful for individualizing medical advice and information to the community.

According to the results of bivariate analyses, the main factors associated with readmission to the ED in our study were polymedication (P = 0.031), living in a residential care home (P = 0.014), fever (P = 0.047), general malaise (P $<$ 0.001), thoracic pain (P $<$ 0.001), headache (P = 0.012), depressive symptoms (P = 0.009), hematological symptoms (P = 0.011), nephrological symptoms (P = 0.047), syncope or hypotension (P = 0.006) and superinfection (P = 0.018). On the other hand, our results point to persistent anosmia or dysgeusia as a possible protective factor for readmission to the ED (P = 0.031). It is possible that these symptoms are more frequent in less severe cases of COVID-19. However, studies specifically aimed at contrasting this hypothesis should be conducted.

We believe that polymedication (more than six drugs per day) and living in residential care homes point to a vulnerable profile of patients that might require strengthening and individualizing post-discharge follow-up strategies. Both factors were confirmed by multivariate analyses (OR: 2.12, 95% CI: 1.16–3.89 and OR: 2.51, 95% CI: 1.16–5.48, respectively).

Regarding symptoms, multivariate analyses confirmed the associations between higher frequency of readmission to the ED and general malaise (OR: 5.24, 95% CI: 1.72–15.94), thoracic pain (OR: 4.45, 95% CI: 1.86–10.63) and hematological symptoms (OR: 3.95, 95% CI: 1.12–13.89). However, given the limitations our study data, headache, depressive symptoms, nephrological symptoms, syncope or hypotension and superinfection should also be cautiously considered as possible markers of severity.

4.3 Limitations of the study

The primary aim of this study was to offer a global description of symptomatology during 6 months after discharge of COVID-19 in our institution. However, we could not gather information on the time of symptom onset, thus it was not possible to distinguish between early or late symptoms, and survival analyses or Cox regression models could not be performed. We used medical records for data gathering. Nevertheless, many potentially relevant variables such as obesity or smoking were not systematically collected in the reports and, therefore, could not be analyzed. We tried to minimize this limitation by collecting all available sociodemographic variables and including them in the multivariate analysis adjustments. It should also be noted that data were obtained from a single institution, thus the external validity of our results should be interpreted cautiously. Finally, possible biases associated with the retrospective nature of this study should be considered. We believe that the detailed information on symptoms and factors associated with readmission to the ED after 6 months of follow-up may contribute significantly to improving future follow-up of COVID-19 patients discharged from hospital.