Predict the hospitalization in COVID-19: Magic is in the air

Background: Although pathogenesis and pattern of disease are still not completely understood, tactical management of overcrowding of hospitals and rational usage of resources is the need of the hour. Aims and Objectives: The study objectives were as follows: Finding of correlation between various attributes of COVID; evaluation of the association of common characteristics with hospital stay; prediction of cooccurrence of different symptoms; calculation of odds ratio of prolonged hospitalization due to various symptoms; and estimation of the rate of prolonged hospitalization due to different symptoms and comorbidities. Materials and Methods: Retrospective data of 6918 COVID-19-positive cases from SCB Medical College and Hospital, India, were obtained from the hospital records from March 2020 to January 2021. The patients’ age, gender, symptoms, and comorbidities were analyzed against their hospital stay using R software (version 4.0.2). Results: Elderly patients (>65 years) had a higher rate (91.22%) of prolonged hospital stay as compared to others (47.61%). Frequently observed symptoms (in decreasing order) were fever (73.93%), cough (67.52%), myalgia (62.11%), dyspnea (49.59%), dizziness (47.38%), and anosmia (44.10%). The risk of prolonged hospitalization was highest with dyspnea [odds ratio: 2.29 (95% confidence interval: 2.07–2.52)], followed by diarrhea [odds ratio [OR] 1.98 (confidence interval [CI] 1.77–2.21)], fever [OR 1.89 (CI 1.69–2.10)], anosmia [OR 1.86 (CI 1.69–2.05)], and dizziness [OR 1.46 (CI 1.32–1.60)]. The rate of hospitalization for more than 7 days was highest with diabetes (86.80%) followed by respiratory illnesses (71.85%) and hypertension (71.28%). Conclusion: These findings can help manage patients based on their symptoms and comorbidities before admission.


INTRODUCTION
Since December 2019, our perceptions, thoughts, and visions have changed because of the domination of coronavirus over our world. Its impact has drastically affected billions of lives around the globe physically, mentally, socially, and most importantly, economically, and the counts are still uprising. The clinical symptoms of COVID-19 infection vary National Journal of Physiology, Pharmacy and Pharmacology 2 2022 | Vol 12 | Issue 04 (Online First) has heightened the necessity of long-term hospital stay, even the need for ventilatory support of severely ill patients. It has alarmed global emergency due to the immediate requirement of massive health infrastructure like hospitals with intensive care unit set up and pharmaceutical companies to combat the nightmare. [5][6][7][8] Prediction of the hospitalization rate is the hour's call for the tactical management of rising disease burden and logical utilization of resources. Hence, we conducted a retrospective study to assess the attributes and predictors of prolonged hospitalization of COVID-19 disease in a tertiary care hospital.

MATERIAL AND METHODS
We performed a retrospective analysis of the duration of hospitalization in 6918 individuals who were tested positive for severe acute respiratory syndrome (SARS)-COVID-19 infection and admitted to SCB Medical College, Cuttack, Odisha, India, within the period of March 25, 2020-January 31, 2021. Individuals of 15-90 years of age, of either gender, with COVID-19 disease admitted to the study site, were included in the study. Exclusion criteria were as follows: Persons hospitalized for less than 24 h; persons referred from and to another hospital; pregnant or lactating women; persons with heart failure, myocardial infection, cerebrovascular accident or thrombo-ischemic attacks, chronic kidney disease (serum creatinine value of more than 1.5 mg/dl), and benign/malignant tumors, burn injury, persons requiring dialysis, persons hospitalized with any bone fractures; persons on total parenteral nutrition or any immunosuppressant drugs; persons who got themselves discharged against the medical advice of doctors; any hospital deaths; and persons reinfected with COVID-19 infection. We got the ethical approval (IEC application no: 601 dated February 11, 2021) from the Institutional Ethics Committee of SCB Medical College and Hospital, Cuttack, Odisha, India, before collecting and analyzing the data. Since the data for this retrospective study were collected from the hospital record, consent for participation in the study was not applicable for this study.
The study objectives were as follows: Finding of a correlation between various attributes of COVID; evaluation of the association of common characteristics with hospital stay; prediction of cooccurrence of different symptoms; calculation of odds ratio of prolonged hospitalization due to various symptoms; and estimation of the rate of prolonged hospitalization due to symptoms and comorbidities. We collected the demographic and clinical data from hospital records with permission from the hospital in charge. A total of 390,732 patients were admitted to the hospital during the stipulated time duration out of which 6918 patients met the inclusion and exclusion criteria of the study and were included for analysis. All the statistical analyses for this study were done using R software (version 4.0.2).  (Figure 1a). Most of the female patients had shorter (<7 days) duration of hospital stay, and male patients had longer (>7 days) hospital stay (Figure 1b). There was an increased duration of hospital stay with increased age and number of symptoms. Patients with bronchial asthma or chronic obstructive pulmonary disease (COPD) were admitted for a longer duration than those without respiratory illnesses (Figure 1c). The duration of hospitalization was prolonged with the presence of respiratory conditions. Supplementary Figure 1a shows that most patients with no respiratory conditions were discharged within 10 days of hospitalization. Patients with bronchial asthma had a longer hospital stay in comparison to COPD cases. There was an inclining trend of the duration of hospitalization with the advancing age of the individual, as shown in Supplementary Figure 1b. Most patients of age <35 years had been discharged within 10 days of admission. All the patients with >5 symptoms had to stay for a longer duration (i.e. more than 7 days) as compared to those with less symptom complex, as shown in Supplementary Figure 1c.

For
The correlation matrix had been plotted in Figure 2. We found significant correlation of duration of hospital stay with following variables: Duration of diabetes (r = 0.78, P < 0.001), age of the person (r = 0.62, P < 0.001), number of symptoms (r = 0.56, P < 0.001), duration of hypertension (r = 0.53, P < 0.001), duration of asthma or COPD (r = 0.52, P < 0.001), and duration of fever before hospitalization (r = 0.44, P < 0.001). Other significant correlations were between following variables: Duration of dyspnea and number of symptoms (r = 0.58, P < 0.001), duration of diabetes and age of the person (r = 0.55, P < 0.001), duration National Journal of Physiology, Pharmacy and Pharmacology 2022 | Vol 12 | Issue 04 (Online First) IQR: Interquartile range of dyspnea and duration of diabetes (r = 0.47, P < 0.001), duration of hypertension and age of the person (r = 0.47, P < 0.001), duration of diabetes and number of symptoms (r = 0.45, P < 0.001), duration of asthma and number of symptoms (r = 0.44, P < 0.001), duration of asthma and age of the person (r = 0.43, P < 0.001), and duration of fever and duration of diabetes (r = 0.42, P < 0.001). There was a single negative (non-significant) correlation of duration of hospital stay with duration of alcoholism (r = −0.07, P = 0.921).
For highlighting the associations among age, respiratory illness, number of symptoms, and duration of hospital stay, a Sankey diagram is drawn in Figure 3. The columns (from left to right) represented the person's age, presence/absence of any respiratory disease, number of symptoms experienced before hospitalization, and duration of hospital stay after diagnosis of infection. The widths of the bands denote the association between the variables. Older adults were more affected by asthma or COPD than younger individuals. The patients with COPD had more symptoms than those with asthma and those without any respiratory diseases. The number of symptoms proportionately affects the duration of hospitalization. All individuals with more than 5 symptoms stayed in the hospital   The up-left plot shows the symptoms in all participants. The up-right portion is meant for the participants hospitalized for 2-7 days. The downleft part depicts the cooccurrence of symptoms for patients hospitalized for more than 7 days. The down-right plot shows participants who were hospitalized for more than 14 days. Fever was the most common symptom experienced among the study participants. It mainly cooccurred with cough, myalgia, dyspnea, and dizziness. Here, the size of the circle indicates the percentage of the participants, and the thickness of the line shows the degree of cooccurrence of the symptoms. The cooccurrence plots suggest the direct relationship of the number of symptoms with the hospital stay, which can otherwise be stated as; more the multi-cooccurrence of symptoms, the longer is the hospital stay. Our study showed the strongest association with the following symptom complexes: Fever-cough (70%) and fever-myalgia (67%). The most petite association was found between loose motions-hoarseness (10%) and dizziness-hoarseness (10%). This finding held good for hospitalization up to 7 days and for >7 days. However, in contrast to the above scenarios, prolonged hospitalization of more than 14 days was strongly associated with multiple symptom complexes, that is, fever-cough, fever-myalgia, fever-breathlessness, and fever-hoarseness. Hence, fever may be considered as the most crucial symptom associated with other symptoms.
In Figure 5, we analyzed the data to estimate the risk of prolonged hospitalization more than 7 days using the symptoms seen in the patients before the diagnosis of COVID-19.  Supplementary Figure 2a and b. These figures suggested that male patients with symptoms such as dizziness, loose motion, dyspnea, and fever were approximately 2 times more likely to stay for >7 days in the hospital. Moreover, female patients with symptoms such as anosmia, hoarseness, dyspnea, and fever were approximately 2-5 times more likely to stay for >7 days in the hospital. In contrast to males, females with dizziness were less likely to stay for a prolonged duration.
We analyzed the data to find the rate of prolonged hospital stay of the study participants with various symptoms, comorbidities, and addiction(s), and the number of signs and created funnel plots shown in Supplementary Figures 3-5. Fever and dizziness are the two prominent symptoms that help predict the rate of hospitalization >7 days, whereas dizziness and hoarseness are the two critical symptoms that indicate the rate of hospitalization >14 days. When the numbers of symptoms are taken into account, a combination of four symptoms comes close to predict prolonged hospital stay (but outside the CI). Hypertension, asthma/COPD, smoking, and tobacco are the significant predictors for hospitalization >7 days, but diabetes replaces as a decisive single factor for hospitalization >14 days.

DISCUSSION
To the best of our knowledge, this is the first study conducted on the correlation of various attributes with the duration of hospital stay of individuals infected with COVID-19 infection in the state of Odisha. Our study provides some insights into the clinical presentation of COVID-19 infection and hospitalization due to the disease. Our study found that symptoms before hospital admission were predictive of the duration of hospital stay. With the number of symptoms before diagnosis, age, and gender, we designed a model to differentiate short (≤7 days) and long (≥14 days) hospitalization due to COVID-19. The model was generalized with the same performance to the population that reported antibody testing.
Elderly individuals, males, smokers, and those with preexisting diabetes and/or hypertension were predominant in this study and the pattern was similar to the study by Richardson et al. [8] Fever was the most common symptom (55.6%), which is consistent with a previous study by Richardson et al. [8] This finding is different from what was reported in a study by Li et al., [9] where cough was found to be the most common symptom. Consistent with the meta-analysis by Yang et al., [10] we found that hypertension and diabetes were the two most common comorbidities in patients with COVID-19. Older patients and those with comorbidities were more likely to experience disease aggravation and require referral to the regular hospital for more intensive care, probably because they had weaker immune systems.
Our study has following strengths. First, we have mentioned the associations of common factors with the duration of hospital stay through the Sankey diagram. This plot depicted that persons with respiratory illnesses, increased age, and number of symptoms were hospitalized for prolonged duration. Second, we have evaluated the cooccurrence of different symptoms of COVID by cooccurrence plots. It was clear from these plots that duration of hospital stay is positively associated with occurrence of multiple symptoms. Third, the funnel plots calculated the hospitalization rate for various symptoms and comorbidities of the hospitalized individual. However, the generalization of this study should not be considered as it has limitations. Our study was limited by including patients diagnosed at, admitted to, and discharged from SCB Medical College, Cuttack, India. Patients not fitting into any one of these were not included for analysis because it would be difficult to trace their clinical records before and after their stay at the study site. Patients below 35 years and above 65 years were underrepresented, which could over or underestimate our duration of hospitalization and rate of prolonged hospitalization. Therefore, for extrapolation of these results into smaller population subgroups, caution is advised. During the early stages, PCR testing was restricted only to those persons who were severely ill, which could have inflated the estimates of the study. Persons with mild symptoms under home isolation, demanding medical supervision, and visiting the study site were tested again with a rapid antigen kit, and those with positive reports were admitted, which could have reduced the accurate estimates of the study as they were mildly infected and therefore discharged sooner than those with multiple, severe symptoms and comorbidities. We excluded patients with heart disease, neurological disorder, and malignancy since these might prolong their hospital stay. Many patients referred from different government health-care centers, private clinics, and hospitals across the state were excluded because the data regarding the onset and severity of symptoms before their diagnosis could not be retrieved. Moreover, literacy, and language were essential barriers to obtain the symptom history and other clinical data regarding the study, mainly for patients from other states and remote areas of the Odisha state, who only knew their vernacular language. All of these could have increased or decreased the estimate of hospitalization. In our study, admission of patients with a wide range and various combinations of symptoms allowed us to state that most people with multiple symptoms and uncontrolled comorbidities are males of the elderly age group. Some studies had not reported bronchial asthma as a risk factor for hospitalization, but its association with prolonged hospitalization demands further investigation.

CONCLUSION
It can be concluded from our study that increasing age, male gender, number, and severity of symptoms, comorbidities such as diabetes, hypertension, bronchial asthma, and COPD are positively associated with duration of hospitalization. This information could help as targeted education material for both affected individuals and health-care providers. Further studies with inclusion of patients with other comorbidities, vaccination status, and decision analysis curve are warranted.