What is the clinical course of patients hospitalised for COVID-19 treatment Ireland: a retrospective cohort study in Dublin’s North Inner City (the ‘Mater 100’)

Setting

The study was conducted at the Mater Misericordiae University Hospital (MMUH). In addition to the local services for the catchment area, the hospital provides a range of frontline and specialist services on a regional and national level, treating 24,750 inpatients, 221,956 outpatients and 82,307 emergency department visits last year. The hospital is located in Dublin’s north inner city with many of Ireland’s most deprived neighbourhoods are situated in the hospital’s catchment area. Reflecting this demographic, infectious diseases care locally has involved addressing communicable diseases such as HIV and hepatitis C in partnership with local communities. Most recently this has involved initiatives involving prisoners¹², homeless populations¹³ and patients attending General Practice¹⁴.

The MMUH Infectious Diseases Department contains the National Isolation Unit (NIU), a six-bed unit with isolation rooms under negative–pressure ventilation, donning and doffing areas of a high specification. During the ‘Containment Phase’ (January to March 2020), our strategy was to contain suspected/confirmed cases of COVID-19 in the NIU. During the ‘Delay Phase’, with rising numbers of cases, the hospital established a ‘COVID pathway’, whereby unscheduled admissions with possible COVID infection were streamed in a parallel system with patients managed on dedicated wards by the team. Once patients were fit for discharge and had appropriate accommodation to self-isolate, they were discharged utilising a mobile health platform (Patient-M-Power®) to monitor their progress.

As in most hospitals in Ireland, from late March, patients were admitted under the care of a ‘COVID Team’, whereby one or all of Infectious Diseases, Respiratory Medicine, Acute Medicine, and Intensive Care specialties, among others, would input to patient care based on the predominant issue requiring clinical intervention.

Sample size

By mid-April, 352 patients had been admitted to the MMUH COVID pathway, 163 of whom received hospital treatment and 189 received outpatient treatment. We carried out a retrospective review of the data of the first 100 patients admitted to the hospital for in-patient treatment of COVID-19 infection from time of disease outbreak in March 2020 to 1^st April 2020.

Subjects

Inclusion criteria: The data of the first 100 adult patients admitted to the COVID pathway with SARS-CoV-2 detected by PCR at the MMUH were eligible for inclusion in this retrospective review.

Exclusion criteria: The data of patients who acquired COVID-19 infection whilst in hospital as an in-patient for treatment for an alternative medical condition were excluded from this study.

Data collection and study instrument

Anonymised data was collected on baseline demographics, clinical parameters and health outcome measures from clinical records (patient charts and computerised patient database) , including: age; gender; type of health insurance; dates of presentation, admission and discharge; presence of pre-existing morbidity or other illnesses at admission; subsequent clinical care; and treatment outcomes. This data was collected from patient records by a member of the clinical team (CC, BOK, SC, EOC, and JL) and entered directly onto an excel database and anonymised to allow for further coding and data analysis by the research team (SC, TMH, GA, and WC).

Data analysis

Statistical analysis was preformed using IBM SPSS Statistics (version 26.0). Frequency counts were presented in respect of categorical and ordinal data, and median / interquartile range (IQR) in respect of numerical data where such data was not normally distributed. We examined patient characteristics (e.g. age, gender, pre-existing morbidity [any], pre-existing cardiorespiratory morbidity¹⁵, GMS status), findings on clinical assessment [e.g. National Early Warning Score (NEWS)¹⁶/NEWS2 Score¹⁷]) and laboratory parameters (e.g. lymphopaenia¹⁸, D-dimer, CRP, ferritin, abnormal chest imaging). For laboratory results, we also assessed whether the measurements were outside the normal range. If it was not possible to find data on clinical parameters this ‘missing data’ was treated in accordance with ‘STROBE Guidance’¹⁹. As such, when reporting proportions for specific variables, the observed finding is presented as a percent of the total number of cases on whom data was available.

Ethical issues

In designing this study we have taken cognisance of best practice in conducting health research during times of major disasters²⁰. Data was collected from clinical and administrative records by a member of the clinical team. All data was anonymised at the time of data collection. All anonymised data was stored as password protected files on a secure server at the University College Dublin Catherine McAuley Research Centre, where it was analysed by the research team. The study was approved by the MMUH Research Ethics Committee (8^th April 2020; reference 1/3782141).

Patient and public involvement

Due to the pandemic nature of COVID-19 it was not feasible to include patient and public involvement. Patients were not invited to comment on any aspect of the study design nor were they invited to contribute to the drafting or editing of this manuscript.

Participants

We present data of the first 100 in-patients admitted to Mater Misericordiae University Hospital for in-patient treatment of COVID-19 infection from time of disease outbreak in March 2020 to 1^st April 2020.

With regards to patient characteristics (see Table 1²¹) 58 (58%) were male, 55 (63%) were Irish nationals, 29 (59%) were GMS eligible and median age was 45 years (IQR=34-64 years). Of these, 83 (88%) were reported as having community-acquired infection; 24 (25%) acquired the infection from household contacts, 13 (14%) from foreign travel, 10 (10%) from their occupation and seven (7%) from working in a healthcare setting. In total, 14 patients worked in healthcare (16%). `

Descriptive data

Patients had symptoms for a median of five days before diagnosis (IQR=2.5-7 days), most commonly cough (72%), fever (65%), dyspnoea (37%), fatigue (28%), myalgia (27%), headache (24%), sore throat (15%), nausea (15%), diarrhoea (11%) and abdominal pain (10%) (See Figure 1).

Figure 1. Presenting symptoms at time of admission.

SOB- shortness of breath.

At least one pre-existing chronic illness was reported for 54 patients, with hypertension (16%), diabetes mellitus (12%), asthma (11%), dyslipidaemia (11%), chronic obstructive pulmonary disease (8%) and a chronic mental health condition (8%) the most commonly observed comorbidities. With regards to presenting features and initial assessment 25% of patients had oxygen saturations <96%, 37% had respiratory rate >24/min, 51% had heart rate >90/min, 32% had temperature >38.0ºC and 14% had systolic BP <111mmHg (see Table 2²¹).

A NEWS score¹⁶ of ≥3 was recorded in 51%, with a NEWS2 score¹⁷ of ≥3 documented in 53% and a modified NEWS²² score ≥3 observed in 55% of patients. With regards to laboratory findings commonly observed abnormalities were: elevated levels of C-Reactive Protein (74%), ferritin (63%) and D-dimer (62%); 42% had a neutrophil / lymphocyte ratio of >3.5, 38% had lymphopaenia; and 15% and 25% had elevated levels of sodium and creatinine, respectively (see Table 3²¹). All 100 had chest imaging; 59 (59%) had an abnormality reported, with bilateral infiltrates documented in 32 (32%) and focal changes in 12 (12%).

Outcomes

Supplemental oxygen was required by 27 patients, of whom 17 were admitted to the intensive care unit - 14 requiring ventilation. Antiviral treatment was administered to 40 patients, most commonly hydroxychloroquine (35%) and lopinavir/ritonavir (16%). Complications documented in the clinical record included: ARDS, diarrhoea / gastrointestinal complaints and acute kidney injury. Four people died and 74 people were discharged home. The median length of hospital stay was nine days (IQR=6-11).

Key results

Our findings reinforce the consensus that COVID-19 is an acute, life threatening disease that is associated with considerable mortality. The study highlights the importance of clinical, laboratory and radiological parameters in assessing disease severity. At admission, the most common abnormalities identified among our cohort were elevated levels of C-reactive protein, ferritin, D-dimer, abnormal chest imaging, a NEWS Score (Modified) of ≥3 and tachycardia.

Correlation with other centres’ experiences

To date, most data on COVID-19 disease has been reported from China although in recent weeks literature has started to emerge from Europe and the United States (Table 4). The baseline clinical characteristics of our population show striking similarities to those reported in studies from China, with regards in particular to male predominance, high prevalence of cough and fever, blood test abnormalities (including lymphopenia and elevated CRP, ferritin and D-dimer levels) and high frequency of abnormalities on chest imaging^10,23–26. Cohorts reported from outside of China (e.g. Korea, Europe and United States) also identified these findings^27–31.

There is little consensus in the published literature to date regarding optimum therapeutic strategies. For example, the use of antibiotics has ranged from 23-95%^10,23–25. Antiviral choice and use is also variable, with one study reporting use of lopinavir/ritonavir (LPV/r) in 21%¹⁰, two studies describing oseltamivir in 90% and 35% of patients, respectively^23,24, and one study reporting ‘antiviral’ administration in 94% without specifying which drugs were used²⁵. The reported rates of systemic corticosteroids has ranged from 3%²⁵ to 45%²³ and the reported rates of use of intravenous immunoglobulin (IVIG) has ranged from 13²⁵–24%¹⁰. The reported rates requiring supplemental oxygenation for COVID-19 infection has ranged from 21%¹⁰ to 90%²⁵ with reported rates of ventilation ranging from 0%²⁵ to 17%¹⁰.

Spiteri et al. describe the first 38 cases in Europe with comparable prevalence of IMV at 9% of hospitalized patients. Since the studies from the United States and Italy (see Table 4) primarily describe patients in an Intensive Care Unit (ICU) setting, the proportion of patients requiring ventilation is higher at 71-88%^29,31. The median length of stay in studies from China has ranged from 10–12 days where such data was available^4,10,24. This is comparable to our cohort who spent a median of nine days in hospital. Furthermore, the rate of complications including requirement for ICU admission, development of ARDS and death also appear comparable to that reported previously.

To date, the largest cohorts have been reported by Guan et al.²⁴, who reported 1099 patients across multiple sites in China and Grasselli et al.²⁹, who reported on 1591 patients admitted to ICUs in Lombardy, Italy. In terms of size, setting and population, the studies which report on a setting most similar to ours are Wang et al.²³ (retrospective study of 138 patients attending Zhongnan Hospital of Wuhan University, China), Yang et al.²⁵ (retrospective study of 149 patients attending three hospitals in Wenzhou, Zhejiang, China) and Caruso et al.³⁰ (single centre prospective cohort study of 158 patients hospitalised in Rome, Italy). Our findings are comparable with these other centres with regard to male gender, age, symptoms at presentation, treatment requirements and laboratory abnormalities. Of our cohort, 75% have been discharged and this is higher than that reported by Wang et al.²³ and Yang et al.²⁵. However, our observed mortality rate of 4% died is the same as that reported by Wang et al.²³. Finally, our cohort characteristics are consistent with those reported in Ireland overall, in terms of age and those most at risk of severe illness being those with an underlying health condition³.

Our findings differ from that reported in other cohorts in that more people had a pre-existing chronic illness. While respiratory symptoms and fever were common among our sample, we also observed less specific symptoms such as myalgia, fatigue and gastrointestinal symptoms (e.g. nausea, diarrhoea) more commonly than has been previously reported.

In our sample, the application (albeit post hoc) of standard ‘early warning scores’^16,17 would have resulted in less than half of people who required admission actually being admitted (45% using NEWS and 48% using NEWS2 parameter, respectively). We therefore recommend against relying on these measures alone when assessing requirement for patient admission, since these scoring systems were developed to aid in decision making for patients with bacterial sepsis as opposed to viral pneumonitis due to COVID-19.

Methodological considerations

This paper reports on real world data from a University teaching hospital in Dublin with a high incidence of COVID-19 disease. As the disease has unfolded it has become apparent that communities living in our local catchment area are especially at risk of the infection and its deleterious consequences³² and this finding has been reported in other large cities³³. The data was collected as our understanding of the natural history of this disease was unfolding. In that regard, while we endeavoured to ensure that the dataset is as complete as possible, this was not always possible and some data points were therefore missing. Nonetheless, we strove to minimise the bias resulting from this by reviewing clinical and administrative records and treating such missing data in accordance with ‘STROBE Guidance’¹⁹.

We acknowledge a number of limitations in our study. Firstly, there may have been a low threshold to admit patients for hospital treatment due to an initial containment strategy and relative absence of capacity constraints at the outbreak’s onset. This may continue to change as both hospital bed capacity and our understanding of the factors associated with worse outcomes evolves in the months ahead. The sample size, while small, is nonetheless, at the time of writing, one of the ten largest cohorts reported to date and to our knowledge the first such from Ireland or the UK.

Implications for clinical practice

COVID-19 is a new disease caused by an emergent virus and is not yet well characterised. Clinical manifestations are different to illness caused by other coronaviridae. Our findings provide a guide on the clinical predictors of patients requiring hospital treatment. The epidemiological characteristics and clinical manifestations of COVID-19 in our cohort appears consistent with findings in cohorts reported to date. Men in their fifth decade appear especially at risk, while elevated levels of C-reactive protein, ferritin and D-dimer, in addition to abnormalities on chest imaging, elevated NEWS Score (Modified) score and tachycardia appear to be important predictors of disease severity. Further research involving larger samples followed longitudinally is a priority and such research will be key to identifying those parameters which best predict disease outcomes and best allocate resources. For clinicians, our key message is that relying exclusively on clinical assessment and tools such as early warning scores alone may not identify those who require hospital care.

Underlying data

Zenodo: What is the clinical course of patients hospitalised for COVID-19 treatment Ireland: a retrospective cohort study in Dublin's North Inner City (the 'Mater 100'). http://doi.org/10.5281/zenodo.4015117²¹

This project contains the following underlying data:

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).