Early experience of an infectious and tropical diseases unit during the coronavirus disease (COVID-19) pandemic, Florence, Italy, February to March 2020

We analysed the first 84 coronavirus disease (COVID-19) patients hospitalised in an infectious and tropical disease unit in Florence, Italy, over 30 days after the start of the COVID-19 outbreak in Italy. A 12% reduction in the rate of intensive care unit transfer was observed after the implementation of intensity care measures in the regular ward such as increasing the nurse/patient ratio, presence of critical care physicians and using high flow nasal cannulae oxygenation.

Since the first case reported on 21 February 2020 in Lombardia region, Italy has become one of the countries most affected by the coronavirus disease (COVID-19) pandemic, counting more than 197,000 cases and more than 26,000 fatalities by 27 April. At the end of April, Italy has the highest numbers of confirmed deaths in Europe [1,2].
We present our first-month experience of the COVID-19 pandemic in Italy at the Infectious and Tropical Disease Unit (ITD) of the University Hospital in Florence, Tuscany. We aimed to describe clinical and demographic characteristics of hospitalised patients with COVID-19 as well as differences between hospitalised patients not transferred to the intensive care unit (ICU) and hospitalised patients later transferred to the ICU. Finally, we analysed the incidence rate of ICU transfers before and after the implementation of additional intensity care measures on our regular ward.

Construction of the cohort and data analysis
We enrolled all laboratory-confirmed COVID-19 cases who were hospitalised in the ITD ward between 25 February and 26 March 2020. A laboratory-confirmed case was a symptomatic patient with positive result on a reverse-transcriptase (RT)-PCR assay. Demographic and clinical data were collected from electronic medical hospital records. ICU transfer was the primary outcome. Categorical variables were analysed by chisquared or Fisher's exact test and continuous variables by the Kruskal-Wallis test. Poisson regression model was used for regression analysis. Cases were followedup from the first day of hospitalisation to ICU transfer or hospital discharge or death. Cases still hospitalised were censored on 3 April. Data collection was approved by the local Ethics Committee (17104_oss). The study was performed in accordance with the ethical principles of the Declaration of Helsinki and with the International Conference on Harmonization Good Clinical Practice guidelines.

Rising intensity of care: pre-and post-intervention
Since the beginning of the COVID-19 outbreak in Italy, the intensity care level in the ITD was gradually increased. Under standard operations, hereafter referred to as 'pre-intervention', our ward has a nurse-patient ratio of 1:10. Thirteen days after hospitalisation of the first COVID-19 case, hereafter referred to as 'post-intervention', the nurse patient ratio was enhanced to 1:6, a stable team of critical care physicians was formed, and advanced ventilation support like high flow nasal cannulae oxygenation became available on the ward.
Additionally, two women, both in the second trimester of pregnancy, were hospitalised.
Corticosteroid treatment was only prescribed in N-ICU patients (n = 14; 20.6%) in the post-intervention period. Overall, three patients received tocilizumab: two in N-ICU and one in ICU. Four patients received remdesivir (5.9%), all in N-ICU.

Clinical and laboratory characteristics
Among laboratory parameters at the admission to the ward, we found a higher median of neutrophils and procalcitonin in N-ICU patients. Moreover, a higher median of C-reactive protein was observed in ICU patients compared with N-ICU (126 mg/L (IQR 72-209) and 44 mg/L (IQR 19-71) respectively; p < 0.0001). Although at the admission interleukin-6 was only available for 50 of 84 patients, we did not observe any difference in the two groups. (Table 2) We did not find any particular variation in clinical presentation between patients transferred or not to ICU, except for diarrhoea which was more common in the N-ICU group (Table 3). Notably, three patients had an atypical presentation without fever: two syncopes, one abdominal pain. Data on self-reported dysgeusia and anosmia were available only for 48 N-ICU of 68 cases, and reported by 54.1% and 35.4%, respectively.

Clinical characteristics in the preintervention and post-intervention period
We observed no difference in age, having COPD, coronary heart disease, chronic hepatitis B, cerebrovascular disease, oncologic illness, chronic renal failure, Horowitz index at admission between COVID-19 cases during the pre-intervention or post-intervention phase. The percentage of patients with diabetes mellitus and hypertension was higher at pre-intervention compared with post-intervention (25.8% vs 7.6% and 58.1% vs 24.5% respectively) ( Table 4). In the pre-and post-intervention period, nine (29.0%) and seven (13.2%) patients were transferred to ICU. The overall rate of ICU transfers in our unit was 2.64 (95% confidence interval (CI): 1.62-4.31) per 100 patient days. The rate of ICU transfers in the postintervention period compared with the pre-intervention period markedly decreased from 6.30 per 100 patient days, (95% CI: 3.00-13.22) to 1.82 patient days, (95% CI: 0.94-3.50). The unadjusted rate ratio (RR) for transfers for the post-intervention period was 0.28 (95% CI: 0.10-0.75; p = 0.0070). After adjustment for sex, age, Horowitz index at baseline, diabetes mellitus, hypertension, COPD, coronary heart disease, the RR for the post-intervention period was 0.88 (95% CI: 0.82-0.95; p = 0.002).

Discussion
The majority of cases hospitalised at ICT during the first 30 days of the COVID-19 pandemic were males older than 60 years of age. More than half of them had at least one comorbidity of which hypertension, coronary heart disease, diabetes mellitus were most frequent, oncologic illness former or still, COPD, chronic renal failure, chronic hepatitis B and cerebrovascular disease were also present in a notable number of patients. We observed that the limited number of cases for which information was available, self-reported olfactory and/or taste disorders in 35% and 54%, which was higher compared with a recent Italian study [3]. It was not possible to precisely estimate if these symptoms preceded or followed those that prompted medical referral. However, anosmia and dysgeusia should be considered as possible clinical indication of COVID-19 and lead to further investigation including testing for infection with severe acute respiratory syndrome coronavirus 2 (SARS CoV-2) [4].
Comparing admission characteristics between ICU and non-ICU cases, we found that COPD and Horowitz index at the time of hospitalisation were associated with ICU transfer in our series of COVID-19 cases ( Table 1). The median days from the onset of symptoms to nasal swab was shorter for cases transferred to ICU probably because of a more severe clinical picture. Concerning laboratory results, we did not notice any difference at admission in IL-6 levels between the two groups, leading us to speculate that the included patients were in an early stage of the disease and IL-6 production might be triggered in the later stages only [5,6].
The overall proportion of cases transferred to ICU described here is higher compared with a Chinese cohort [7]. However, in our preliminary analysis, we observed a 12% reduction in ICU transfer rate adjusted for comorbidities between the pre-and post-intervention period (p = 0.002). This benefit is represented by avoiding endotracheal intubation, taking out possible infectious risks. However, endotracheal intubation should not be delayed in patients with haemodynamic instability, multiorgan failure, or abnormal mental status [8]. Unfortunately, we could not now evaluate if the intervention had an impact also on mortality, because 19 (22.6%) patients were still hospitalised, and mortality rate could have been underestimated in N-ICU. Current evidence against corticosteroid application in COVID-19 is unclear, but the Chinese Thoracic Society recently developed an expert consensus statement on the use of corticosteroids SARS-CoV-2 pneumonia [9]. The increase in corticosteroid prescriptions in the postintervention period could have played a role in reducing ICU transmission rate. Because of the small sample size and limited outcome information, we could not run a solid multivariable analysis and these results should be cautiously interpreted.
In conclusion, in this Italian cohort, the majority were men older than 60 years with at least one comorbidity in nearly 60%. A self-reported anosmia and/or dysgeusia were frequent symptoms similar to what has been described in a recent European and Italian study [3,4]. The rising intensity of care seemed to reduce the ICU transfer rate and endotracheal intubation. The main limitation of the this study is the small sample size analysed.