Mortality rates among COVID-19 patients hospitalised during the first three waves of the epidemic in Milan, Italy: A prospective observational study.

INTRODUCTION: This paper describes how mortality among hospitalised COVID-19 patients changed during the first three waves of the epidemic in Italy.
METHODS: This prospective cohort study used the Kaplan-Meier method to analyse the time-dependent probability of death of all of the patients admitted to a COVID-19 referral centre in Milan, Italy, during the three consecutive periods of: 21 February-31 July 2020 (first wave, W1), 1 August 2020-31 January 2021 (second wave, W2), and 1 February-30 April 2021 (third wave, W3). Cox models were used to examine the association between death and the period of admission after adjusting for age, biological sex, the time from symptom onset to admission, disease severity upon admission, obesity, and the comorbidity burden.
RESULTS: Of the 2,023 COVID-19 patients admitted to our hospital during the study period, 553 (27.3%) were admitted during W1, 838 (41.5%) during W2, and 632 (31.2%) during W3. The crude mortality rate during W1, W2 and W3 was respectively 21.3%, 23.7% and 15.8%. After adjusting for potential confounders, hospitalisation during W2 or W3 was independently associated with a significantly lower risk of death than hospitalisation during W1 (adjusted hazard ratios [AHRs]: 0.75, 95% confidence interval [CI] 0.59-0.95, and 0.58, 95% CI 0.44-0.77). Among the patients aged >75 years, there was no significant difference in the probability of death during the three waves (AHRs during W2 and W3 vs W1: 0.93, 95% CI 0.65-1.33, and 0.88, 95% CI 0.59-1.32), whereas those presenting with critical disease during W2 and W3 were at significantly lower risk of dying than those admitted during W1 (AHRs 0.61, 95% CI 0.43-0.88, and 0.44, 95% CI 0.28-0.70).
CONCLUSIONS: Hospitalisation during W2 and W3 was associated with a reduced risk of COVID-19 death in comparison with W1, but there was no difference in survival probability in patients aged >75 years.

1. Introduction

Almost two years after the start of the coronavirus disease 19 (COVID-19) pandemic caused by severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), more than 326,279,424 confirmed infections and 5,536,609 COVID-19 related deaths had been reported worldwide [1]. Although most SARS-CoV-2 infections were asymptomatic or only mildly or moderately symptomatic and did not require hospitalisation, hospitalisation rates ranged from 13 to 21 cases per 1,000 confirmed infections, and mainly depends on the age of the infected population [2, 3].

One of the concerning aspects of the pandemic is the probability of death among diagnosed subjects: initial analyses of data from nine countries (China, France, Germany, Italy, The Netherlands, South Korea, Spain, Switzerland and the United States) made in April 2020 showed wide variations in the overall case fatality rate (CFR), which ranged from 0.7% in Germany to 9.3% in Italy, with two-thirds of this variability being explained by differences in the age distribution of the cases [4]. In Italy, the CFR had increased to 14% by June 2020, mainly because of increasing age-specific case fatality [5]. Moreover, the first observational studies in China, Italy and the United States indicated mortality rates ranging from 12% to 28% among patients hospitalised with COVID-19 and going as high as 49% among those admitted to an intensive care unit (ICU), with the highest rates being observed among older patients and those with pre-existing comorbidities [6-8]. However, although the pandemic has continued in waves in many parts of the world and is still keeping health systems under pressure, some European and American studies have shown that the in-hospital mortality associated with COVID-19 has progressively declined [9-13]. This has been variously attributed to rapid changes in hospital organisation that allowed the better management of patient surges (i.e. expanded COVID-19 wards, increased supplies of ventilators and other critical equipment); earlier hospitalisation as a result of easier access to testing; and improvements in COVID-19 treatment [9-13], but it is also possible that the increasing circulation of SARS-CoV-2 and the implementation of measures to shield the most vulnerable populations contributed by changing the case-mix of hospitalised patients towards those who are younger and at lower risk of dying [14].

Since the arrival of the pandemic in late February 2020, Italy has experienced three major surges of COVID cases and deaths, and a fourth wave of contagion is currently ongoing. The cumulative number of COVID-19 deaths as of 9 January 2022 was 137,705 [15], with the highest number being recorded in Lombardy [16], the region that was the epicentre of the first outbreak in the country [16]. During the early days of the COVID-19 epidemic, a registry of all of the patients admitted to the Department of Infectious Diseases and the ICU of Milan’s Luigi Sacco Hospital was established in order to provide initial information concerning the characteristics of the disease and its lethality [17-19].

The aim of this study of updated registry data is to describe how mortality changed during the different phases of the COVID epidemic in Italy and consider the potential drivers of the changes.

2. Materials and methods

2.1 Study design and setting

This prospective cohort study enrolled all of the adult COVID-19 patients admitted to the Department of Infectious Diseases and the ICU of Luigi Sacco Hospital. The characteristics of the hospital and the rapid reorganisation require in order to respond to the pandemic have been previously described [17-19].

2.2 Participants

The analysis considered all of the adult COVID-19 patients admitted to the hospital’s Department of Infectious Diseases and ICU between 21 February (the day the first patients were hospitalised) and 30 April 2021, who were observed until the time of death or censored three months after hospital admission, whichever came first. The diagnosis of COVID-19 was confirmed by a positive real-time reverse-transcription polymerase chain reaction on a nasopharyngeal swab (NPS).

2.3 Data collection

The characteristics of our registry data management have been described elsewhere [17-19]. In brief, the data were extracted from the patients’ clinical charts on a daily basis and stored in an ad hoc database. The collected data were the patients’ date and place of birth, and biological sex; the date of admission, and the time between symptom onset and hospital admission; comorbidities (including diabetes, lung diseases, heart diseases, renal diseases, immune system diseases, liver diseases, and obesity, defined as a body mass index of ≥30) [20]; the burden of comorbidities (0, 1, 2, and 3+); disease severity upon hospital admission (defined as mild, moderate, severe or critical in accordance with the WHO guidelines for the management of COVID-19) [21]; the type of supportive oxygen therapy upon hospital admission and/or during the hospital stay; the drugs used to treat COVID-19 (which included hydroxychloroquine, lopinavir/ritonavir, remdesivir, tocilizumab and other immunomodulators, heparin, and steroids); and the hospitalisation outcome (death, discharge, or transfer to other facilities).

The vital status of the patients who were discharged before the censoring date was ascertained by means of telephone calls and, when appropriate, during an on-site examination by a member of our dedicated post-COVID-19 outpatient service.

For the purposes of this study, the patients were divided into three groups on the basis of the time of their hospital admission. The three consecutive periods were characterised by surges in COVID-19 cases and hospitalisations in Italy: 21 February-31 July 2020 (first wave, W1); 1 August 2020–31 January 2021 (second wave, W2); and 1 February-30 April 2021 (third wave, W3) (S1 Fig) [15–16, 22].

2.4 Outcomes

The main outcome of interest was death.

2.5 Data analysis

The descriptive statistics show proportions for categorical variables, and median values with their interquartile range (IQR) for continuous variables. The baseline demographic and clinico-epidemiological characteristics of the patients admitted during W1, W2 and W3 were compared using the χ2 or, when necessary, Fisher’s exact test in the case of categorical variables, and Wilcoxon’s rank-sum test in the case of continuous variables.

The time-dependent probability of death among the patients admitted during W1, W 2 and W3 was assessed using the Kaplan-Meier method, with Dunn-Sidak adjustment for multiple comparisons. Univariable Cox proportional hazard models were used to investigate the relationship between the period of admission (W1, W2 or W3) and the outcome of interest (death). Associations were estimated using hazard ratios (HRs) with their 95% confidence intervals (CIs). Multivariate Cox proportional hazard models were used to adjust the HRs for potential confounders including age, biological sex, the time from symptom onset to admission, disease severity upon admission, the co-morbidity burden, and obesity, and the results were expressed as adjusted hazard ratios (AHRs) with their 95% CIs.

It was hypothesised that the burden of the spread of SARS-CoV-2 in the province of Milan may have influenced mortality rates in the different study periods as it was presumed that a larger number of infections in the general population would be associated with a larger number of severely or critically ill patients being hospitalised and greater difficulties in their management. Consequently, a variable in which each hospital admission date was attributed with the monthly number of SARS-CoV-2 infections recorded in the province of Milan [22] was created and included in the multivariable model as a possible confounder of the risk of death during W2 and W3 (W1 was excluded because the shortage of reagents for NPS testing and the policy of restricting testing to symptomatic patients affected the reliability of any estimate of the number of infected subjects during this period).

Furthermore, a sub-analysis of the risk of death was carried out by stratifying the patients on the basis of their age (<46 years; 46–60 years; 61–75 years; >75 years) and disease severity upon hospital admission (the WHO categories of mild/moderate; severe; critical). We also explored changes in survival probability across the waves in subgroups of patients stratified by multiple variables including sex, age, clinical conditions at hospital admission, and comorbidity burden.

The study was approved by our Comitato Etico Interaziendale Area 1 (Protocol No. 16088). Informed consent was waived in the case of patients undergoing mechanical ventilation upon admission.

3. Results

Of the 2,023 COVID-19 patients admitted to our centre during the study period, 553 (27.3%) were admitted during W1, 838 (41.5%) during W2, and 632 (31.2%) during W3.

3.1 Admission characteristics

Table 1 shows the distribution of the patients’ characteristics upon admission as a whole and by wave.

Table 1

Characteristics of the study population as a whole and by period of hospital admission.

		Wave
	Overall n = 2023	1 (21 Feb—31 Jul 2020) n = 553	2 (1 Aug 2020–31 Jan 2021) n = 838	3 (1 Feb– 30 Apr 2021) n = 632
Biological sex, n (%)
Female	722 (35.7)	185 (33.5)	300 (35.8)	237 (37.5)
Male	1301 (64.3)	368 (66.5)	538 (64.2)	395 (62.5)
Median age, years [IQR]	65.7 [54.8, 76.9]	60.8 [49.3, 72.6]	68.5 [56, 80.1]	66.4 [56.9, 75.8]
Age, n (%)
<46	225 (11.1)	103 (18.6)	78 (9.3)	44 (7.0)
46–60	540 (26.7)	161 (29.1)	205 (24.5)	174 (27.5)
61–75	672 (33.2)	175 (31.6)	251 (30.0)	246 (38.9)
>75	586 (29.0)	114 (20.6)	304 (36.3)	168 (26.6)
Italian nationality, n (%)	1633 (80.7)	446 (80.7)	648 (77.3)	539 (85.3)
Comorbidities, n (%)
Obesity‡,	488 (24.1)	98 (17.7)	195 (23.3)	195 (30.9)
Diabetes	315 (15.6)	62 (11.2)	137 (16.3)	116 (18.4)
Lung disease	301 (14.9)	86 (15.6)	144 (17.2)	71 (11.2)
Heart disease	1124 (55.6)	268 (48.5)	487 (58.1)	369 (58.4)
Renal disease	170 (8.4)	45 (8.1)	86 (10.3)	39 (6.2)
Oncological disease	252 (12.5)	55 (9.9)	126 (15.0)	71 (11.2)
Immune system disease	149 (7.4)	43 (7.8)	67 (8.0)	39 (6.2)
Liver disease	71 (3.5)	12 (2.2)	39 (4.7)	20 (3.2)
No. of comorbidities, n (%)
0	532 (26.3)	173 (31.3)	195 (23.3)	164 (25.9)
1	661 (32.7)	183 (33.1)	261 (31.1)	217 (34.3)
2	524 (25.9)	128 (23.1)	224 (26.7)	172 (27.2)
≥3	306 (15.1)	69 (12.5)	158 (18.9)	79 (12.5)
Median time from symptom onset, [IQR]	7 [4, 10]	7 [4, 10]	6 [3, 9]	8 [5, 11]
O2 therapy upon admission, n (%)
None	501 (24.8)	164 (29.7)	243 (29.0)	94 (14.9)
Nasal cannula	511 (25.3)	126 (22.8)	181 (21.6)	204 (32.3)
Venturi	440 (21.7)	88 (15.9)	178 (21.2)	174 (27.5)
Reservoir	136 (6.7)	36 (6.5)	51 (6.1)	49 (7.8)
C-PAP	369 (18.2)	113 (20.4)	158 (18.9)	98 (15.5)
Mechanical ventilation	66 (3.3)	26 (4.7)	27 (3.2)	13 (2.1)
Disease severity upon admission † , n (%)
Mild	167 (8.3)	49 (8.9)	89 (10.6)	29 (4.6)
Moderate	852 (42.1)	241 (43.6)	338 (40.3)	273 (43.2)
Severe	569 (28.1)	124 (22.4)	226 (27.0)	219 (34.7)
Critical	435 (21.5)	139 (25.1)	185 (22.1)	111 (17.6)

Abbreviations: n: number; IQR: interquartile range.

‡Defined as a body mass index of ≥30 [20]

†According to the WHO classification [21]

Most of the patients were males (64.3%), and there was no between-wave difference in the distribution of their biological sex. The patients admitted during W1 were younger than those admitted during W2 or W3 (60.8 years [IQR 49.3–72.6] vs 68.5 years [IQR 56–80.1] and 66.4 years [IQR 56.9–75.8]), and patients aged >75 years were more represented in W2 than W1 or W3 (36.3% vs 20.6% and 26.6%). The patients admitted during W1 were also more frequently free of comorbidities than those admitted during W2 or W3 (31.3% vs 23.3% and 25.9%). The median time from symptom onset to hospitalisation was shorter among the patients admitted during W2 than among those admitted during W1 or W3 (6 days [IQR 3–9] vs 7 days [IQR 4–10] and 8 days [IQR 5–11]). The number of patients requiring immediate oxygen support was lower during W3 than during W1 or W2 (14.9% vs 29.7% and 29.0%), but the overall proportion of patients presenting with critical disease was higher during W1 than during W2 or W3 (25.1% vs 22.1% vs 17.6%).

3.2 Oxygen support, treatments, and outcomes

Table 2 shows the type of oxygen support and pharmacological intervention required during hospitalisation.

Table 2

Characteristics of oxygen support, pharmacological treatments, and outcomes.

		Wave
	Overall n = 2023	1 (21 Feb– 31 Jul 2020) n = 553	2 (1 Aug 2020–31 Jan 2021) n = 838	3 (1 Feb– 30 Apr 2021) n = 632
Advanced respiratory support
C-PCP, n (%)	747 (36.9)	202 (36.5)	303 (36.2)	242 (38.3)
Invasive ventilation, n (%)	316 (15.6)	88 (15.9)	144 (17.2)	84 (13.3)
Maximum level of O2 support, n (%)
No oxygen support	314 (15.5)	117 (21.2)	169 (20.2)	28 (4.4)
Nasal cannula	391 (19.3)	94 (17.0)	140 (16.7)	157 (24.8)
Venturi	369 (18.2)	76 (13.7)	142 (16.9)	151 (23.9)
Reservoir	139 (6.9)	39 (7.1)	59 (7.0)	41 (6.5)
C-PAP	494 (24.4)	139 (25.1)	184 (22.0)	171 (27.1)
Invasive ventilation	316 (15.6)	88 (15.9)	144 (17.2)	84 (13.3)
Pharmacological treatment, n (%)
Hydroxychloroquine	418 (20.7)	417 (75.4)	0 (0.0)	1 (0.2)
Lopinavir/ritonavir	326 (16.1)	326 (59.0)	0 (0.0)	0 (0.0)
Remdesivir	391 (19.3)	87 (15.7)	141 (16.8)	163 (25.8)
Tocilizumab	123 (6.1)	116 (21.0)	1 (0.1)	6 (0.9)
Heparin	1517 (75.0)	202 (36.5)	731 (87.2)	584 (92.4)
Steroids	1281 (63.3)	64 (11.6)	649 (77.4)	568 (89.9)
Median number of days of hospitalisation [IQR]	13 [8, 23]	12 [8, 22]	15 [9, 25]	12 [8, 22]
Deaths, n (%)	417 (20.6)	118 (21.3)	199 (23.7)	100 (15.8)

Abbreviations: n: number; IQR: interquartile range.

Advanced respiratory support with C-PAP was required by 36.9% of patients, and invasive mechanical ventilation by 15.6%, and there was no between-wave difference in the proportion of patients who underwent either. Most of the patients (84.5%) received some oxygen support; the percentage not requiring it was lower during W3 than during W1 or W2 (4.4% vs 21.2% and 20.2%).

On the basis of the evolving evidence provided by clinical trials during the study period, the use of remdesivir, heparin and steroids progressively increased from 15.7%, 36.5% and 11.6% during W1 to 25.8%, 92.4% and 89.9% during W3.

Overall, the median duration of hospitalisation was 13 days (IQR 8–23), but it was longer during W2 (15 days, IQR: 9–25) than during W1 or W3 (12 days, IQR: 8–22 in both periods).

Four hundred and seventeen of the study patients died (20.6%). The crude mortality rate during W1, W2 and W3 was respectively 21.3%, 23.7% and 15.8%.

3.3 Survival analysis and the factors associated with the risk of COVID-19-related death

Fig 1 shows the time-dependent probability of survival during the different waves of the epidemic.

Fig 1

Time-dependent survival probability by period of hospital admission: Wave 1 (February-July 2020), wave 2 (August 2020-January 2021), and wave 3 (February-April 2021).

The probability of death was lower among the patients admitted during the third wave of the epidemic than during the first or second wave. The probability of being alive 30 days after hospital admission was 80.2% (95% CI 76.9–83.6%) during W1, 79.3% (95% CI 76.5–82%) during W2, and 85.8% (95% CI 83–88.5%) during W3.

Table 3 shows the results of the uni- and multivariable analyses of the effect of the period of hospital admission on the risk of COVID-19-related death. Being hospitalised with COVID-19 during W2 or W3 was independently associated with a significantly lower risk of death in comparison with being hospitalised during W1 (AHR 0.75, 95% CI 0.59–0.95, and AHR 0.58, 95% CI 0.44–0.77).

Table 3

Cox model of the effect of the period of hospital admission on the risk of COVID-19-related death.

	HR	HR 95% CI	AHR	AHR 95% CI
Age (per each additional 10 years)	1.80	1.67–1.94	2.18	1.99–2.39
Time from symptom onset (per each additional day)	0.99	0.98–1.01	0.99	0.98–1.01
SARS-CoV-2 epidemic wave 2 vs 1	1.11	0.88–1.39	0.75	0.59–0.95
SARS-CoV-2 epidemic wave 3 vs 1	0.71	0.55–0.93	0.58	0.44–0.77
Males vs females	1.01	0.90–1.34	1.36	1.09–1.68
Obesity: yes vs no	1.22	0.99–1.52	1.66	1.32–2.07
No. of comorbidities ≥3 vs <3	1.58	1.25–1.99	1.00	0.79–1.28
Disease severity: severe vs mild/moderate	2.19	1.71–2.80	2.20	1.70–2.83
Disease severity: critical vs mild/moderate	3.79	2.99–4.80	5.03	3.90–6.49

Abbreviations: HR: hazard ratio; AHR: adjusted hazard ratio; CI: confidence interval.

The multivariable analyses also confirmed that an older age (AHR 2.18, 95% CI 1.99–2.39 per each additional 10 years), male sex (AHR 1.36, 95% CI 1.09–1.68), obesity (AHR 1.66, 95% CI 1.32–2.07), and disease severity (AHR 2.20 (95% CI 1.70–2.83 for severe vs mild/moderate disease, and 5.03, 95% CI 3.90–6.49 for critical vs mild/moderate disease) were all independently associated with a higher risk of COVID-19-related death.

A sensitivity analysis of the risk of death restricted to patients admitted during W2 and W3 showed a significantly lower hazard ratio among the latter (HR 0.64, 95% CI 0.51–0.82). However, this was not confirmed when the analysis was further adjusted for the monthly number of cases diagnosed in the province of Milan during the two epidemic waves (AHR W3 vs W2 = 0.85, 95% CI 0.65–1.11) (S1 Table).

3.4 Survival analysis by age strata and clinical condition at admission

Fig 2 shows the survival curves relating to the three epidemic waves by age strata. There was no difference between the survival curves of the patients aged >75 years during the three periods and, after adjusting for confounders, no significant difference in the risk of death between the patients admitted during W2 or W3 and those admitted during W1 was observed (AHRs 0.93, 95% CI 0.65–1.33, and 0.88, 95% CI 0.59–1.32).

Fig 2

Time-dependent survival probability by period of hospital admission (wave 1 [February-July 2020], wave 2 [August 2020-January 2021], and wave 3 [February-April 2021]) stratified by age: A ≤45 years; B 46–60 years; C 61–75; and D >75 years.

However, the survival probability of the patients aged 46–60 years was different between waves: after adjusting for potential confounders, being admitted during W2 and W3 was independently associated with a significantly reduced risk of death (AHRs of 0.42 [95% CI 0.19–0.94] and 0.17 [95% CI 0.05–0.61]). The survival probability of the patients aged 61–75 years was also different between waves: the adjusted multivariable model showed that the risk of dying was significantly lower during W3 than during W1 (AHR 0.51 [95% CI 0.33–0.78]).

Fig 3 shows the survival curves by disease severity upon hospital admission during the different study periods.

Fig 3

Time-dependent survival probability by period of hospital admission (wave 1 [February-July 2020], wave 2 [August 2020-January 2021], and wave 3 [February-April 2021]) stratified by disease severity upon hospital admission.

A: mild/moderate disease; B: severe disease; C: critical disease.

There was no difference in survival between the patients admitted with severe disease during the three periods: after adjusting for confounders, the estimated risk of death of the patients admitted during W2 or W3 was not significantly different from that of the patients admitted during W1 (AHRs of 0.95 [95% CI 0.59–1.54] and 0.85 [95% CI 0.51–1.40]). However, the survival probability of the patients admitted with mild/moderate disease was different between waves: after adjusting for potential confounders, the patients admitted during W3 were at significantly lower risk of dying than those admitted during W1 (AHR 0.57 [95% CI 0.33–0.97]). Finally, the survival probability of the patients presenting with critical disease was also different between waves: after adjusting for potential confounders, the patients with critical disease admitted during W2 or W3 were at significantly lower risk of dying than those admitted during W1 (AHRs of 0.61 [95% CI 0.43–0.88] and 0.44 [95% CI 0.28–0.70]).

S2 Fig shows the results of the survival analysis by period of admission in subgroups of patients stratified by multiple characteristics. There was an overall improvement in survival in the subgroups aged ≤75 years, which was more marked among males with critical disease (W2 HR 0.57 [95%CI 0.36–0.91] and W3 HR 0.31 [95%CI 0.17–0.56] vs W1), ≥3 comorbidities (W2 HR 0.47 [95%CI 0.19–1.13] and W3 HR 0.37 [95%CI 1.37] vs W1) and those with <3 comorbidities (W2 HR 0.75 [95%CI 0.50–1.10] and W3 HR 0.45 [95%CI 0.28–0.71] vs W1); however, there was no improvement in survival in any of the subgroups aged >75 years.

4. Discussion

The mortality rate among the COVID-19 patients hospitalised at our clinical centre in Milan was lower (15.8%) during the third wave of the epidemic (February-April 2021) than during the first (February-July 2020: 21.3%) or second wave (August 2020-January 2021: 23.7%). It is interesting to note that a more rapid decrease in in-hospital mortality has been observed in some European studies (as early as during the first or second wave), and that this paralleled an increase in the proportion of women and/or younger patients, who are known to be at lower risk of dying [9-13]. However, when the risk of death of our patients admitted during the different waves of the epidemic was adjusted for the known confounders of age, sex, comorbidities and the severity of COVID-19, we found that those hospitalised during W2 or W3 respectively had a 25% and 42% lower risk of dying than those admitted during W1.

Males accounted for 64.3% of our study population and remained over-represented throughout the three waves, which is in line with other national and international findings [23, 24]. Furthermore, our observation that the risk of death was 36% higher among males than among females is in line with the findings of the OpenSAFELY study, one of the largest cohort studies of the factors associated with the risk of COVID-19-related death, which showed that male gender was associated with a 59% excess risk of mortality [24].

The median age of our COVID-19 patients as a whole was 65.7 years, with 62.2% of the patients being aged >60 years. Patients in the age group of 61–75 years were the most represented during W1 and W3, whereas patients aged >75 years were the most represented during W2, which was also characterised by a higher prevalence of patients with multiple comorbidities. It is well known that an older age and a high morbidity burden are closely associated with an increased risk of COVID-19-related death [25], and so it is likely that these factors together were the major drivers of the greater crude mortality observed during W2.

It is worth noting that, despite the recognised importance of implementing integrated care programmes aimed at protecting the most vulnerable against COVID-19 while mitigating the adverse effects of social isolation [26], the age pattern of the patients enrolled in our study underlines the challenges of ensuring the adequacy of such plans during periods of intense disease recrudescence. This has also been highlighted by other single-centre experiences, and national surveillance data show a peak of COVID-19 cases, hospitalisations and deaths in Italian long-term facilities and nursing homes during October-November 2020, corresponding to W2 [27, 28].

Our observation of a significant reduction in the proportion of patients aged >75 years during W3 may indicate an initial effect of the Italian COVID-19 vaccination campaign, which started in late December 2020, was primarily targeted at-risk populations (including the elderly) and, as of March 2021 [29], had led to 49.4% coverage among people aged >80 years living in Lombardy.

The results of our multivariable analysis of the risk of death in stratified age groups showed that it significantly decreased in patients aged 46–60 and 61–75 years between W1 and W3, but not in those aged >75 years. A similar figure was observed when patients were stratified into subgroups according to age and other relevant clinical characteristics (sex, disease severity and comorbidity burden). This suggests that the natural history of COVID-19 in elderly patients is not affected by the current means of intervention and, once again, underlines the pivotal role of preventive measures and vaccination in this population [30].

In line with the findings of two previous Italian studies [31, 32], we also observed a trend towards a reduction in the proportion of critically ill patients upon admission between W1 and W3 (from 25.1% to 22.1% and then 17.6%). This change may have been due to increased access to COVID-19 testing and the detection of cases generally associated with a lower risk of death (testing was initially restricted to the most severely ill patients). The results of our multivariable analysis of the risk of death by disease severity upon admission showed a significant decrease in the risk among the patients admitted with critical disease during W2 (a 38% lower risk) and W3 (a 56% lower risk), which may have been due to rapid improvements in respiratory support and pharmacological management.

The proportion of patients who eventually required invasive mechanical ventilation during hospitalisation was highest during W2, but it must be acknowledged that severe resource constraints during W1 may have prevented a number of patients (particularly frailer patients with severe comorbidities) from gaining access to invasive mechanical ventilation [33]. Moreover, the smaller number of SARS-CoV-2 infections recorded in the province of Milan during W3 [22] may have led to a lower proportion of patients with the most critical disease phenotype requiring intensive care.

As noted by others, the significant change in COVID-19 treatment following the first wave of the epidemic reflected an increasing understanding of its underlying pathogenetic mechanisms (including coagulopathy and hyper-inflammation) and the evidence coming from clinical trials. The use of more effective pharmacological strategies such as thromboprophylaxis [34] and the appropriate use of corticosteroids [35] and remdesivir [36], combined with an overall improvement in respiratory management [37], may therefore explain much of the reduction in the risk of death observed after W1. However, our finding of a further reduction between W2 and W3 is more difficult to interpret because there were no significant changes in patient management between the two periods, and the cause may lie outside the investigated potential confounders. The results of our sensitivity analysis of the possible effect of the intensity of the wave on the risk of in-hospital death suggest that the reduction during W3 may have been at least partially driven by the lower incidence of COVID-19 in Lombardy [22].

It is unlikely that changes in the virulence of SARS-CoV-2 played a role in the decreasing mortality rate over time because, during W2 and W3, there was an increase in the circulation of the alpha variant (B.1.1.7) in Italy [38, 39], and some studies in the UK have shown that this variant is associated with an increased risk of needing critical care and greater mortality than non-B.1.1.7 viral strains [40]. It can also be excluded that COVID-19 vaccination contributed to the decreased mortality observed during W3 because, although W3 overlapped with the early phase of the Italian vaccination campaign, none of our patients was fully vaccinated against COVID-19, and only two had received one dose of vaccine. On the other hand, it is still important to continue monitoring the case fatality rates associated with COVID-19 and the impact that emerging viral variants and vaccination coverage have on the risk of hospitalisation and COVID-19-related death.

This study has a number of limitations. Firstly, it was a single-centre study so its findings may not be generalisable to other centres in Italy or other countries. However, we think that it is important to continue to collect information from various clinical contexts in order to improve our understanding of how in-hospital mortality evolves in different health systems and patient populations. Secondly, the criteria we used to define the different waves of the epidemic was arbitrary and simply based on the shape of the surges and declines in the number of new COVID-19 cases recorded in Italy during the study period; recently proposed mathematical methods of partitioning time series may better define waves and mortality rates during waves [41]. Thirdly, our dataset did not include data concerning the patients’ occupations, socio-economic status, household income, or residence in collective settings such as nursing homes, all of which may contribute to making some groups particularly susceptible to worst health outcomes. Fourthly, although we tried to adjust for the number of SARS-CoV-2 diagnoses in the province of Milan, it is possible that other unmeasured confounders have been missed and that the patients hospitalised during the three waves did not have the same a priori probability of death regardless of medical interventions. Finally, a number of structural and human resource-related changes were made during the study period as the pandemic evolved, and these may have affected our findings by introducing biases that could not be accounted for in the analyses.

5. Conclusions

The second and third waves of the COVID-19 epidemic were respectively associated with a 25% and 42% lower risk of death than the first wave, and this is probably related to an improvement in the overall management of COVID-19 patients. The patients who received the greatest benefit were those aged 46–60 years and those who were critically ill upon hospital admission, whereas in-hospital mortality among the patients aged >75 years remained unchanged throughout the study period. The dynamic nature of the SARS-CoV-2 pandemic, with the evolution of new variants of concern and their possible effects on the success of the vaccination campaign, suggests the need for further evaluations of the management of COVID-19 patients, the most vulnerable of whom have not changed since the virus first arrived in Italy.

A graphical representation of the number of monthly admissions for COVID-19 during the study period (red bars) and the monthly number of new SARS-CoV-2 diagnoses in the province of Milan (black line).

(TIF)

Click here for additional data file.

Time-dependent survival probability in different subgroups by period of hospital admission (wave 1 [February-July 2020], wave 2 [August 2020-January 2021], and wave 3 [February-April 2021]).

1) A,B,C,D being male or female and aged >75 or ≤75 years; 2) A,B,C,D being male or female, aged >75 or ≤75 years, and presenting with critical disease upon hospital admission; 3) A,B,C,D being male or female, aged >75 or ≤75 years, and presenting without critical disease upon hospital admission; 4) A,B,C,D being male or female, aged >75 or ≤75 years, and with ≥3 co-morbidities; 5) A,B,C,D being male or female, aged >75 or ≤75 years, and presenting with <3 co-morbidities.

(PDF)

Click here for additional data file.

Cox model of the factors associated with death during the second and third waves of the epidemic waves, adjusted for the monthly number of new SARS-CoV-2 diagnoses in the province of Milan.

(DOCX)

Click here for additional data file.

(TXT)

Click here for additional data file.

21 Feb 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Chiara Lazzeri

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This paper is very well written indeed. Almost no typos at all can be found. The paper is pleasant and easy to read. The paper has a robust discussion section, including appropriate limitations. The results are carefully presented and the statistical analysis is sound.

The one limitation that isn’t stated is that any choice of “waves” to divide the pandemic into is somewhat arbitrary. There are more mathematical ways to determine such waves rather than simply by inspection, as proposed in this paper “COVID-19 second wave mortality in Europe and the United States” (https://pubmed.ncbi.nlm.nih.gov/33810707/). It could probably be cited regardless as its subject matter is very similar to yours (reduction of mortality in subsequent Covid waves).

Apart from that, there are some truly minor typos. They may have been introduced in my version of MS Word, perhaps.

co-morbidities->comorbidities

up-dated->updated

re-organisation->reorganisation

Reviewer #2: Manuscript N° PONE-D-22-01769 “Mortality rates among COVID 19 patients hospitalized during the first three waves of the epidemic in Milan, Italy: a prospective observational study”

This is an interesting study that analyzes relevant issues in the history of SARS-CoV-2 infection and COVID-19 disease in a specific and highly relevant Italian hospital. It assessed if and how COVID-19 mortality differed between three consecutive periods in which the knowledge and consequently the management of the disease improved considerably.

However, I see a few major methodological issues and other points in the manuscript that need to be addressed in my opinion. They are as follows:

- I strongly recommend the authors to dismiss statistical significance/null hypothesis testing in the entire study, including data analysis and interpretation. They should instead focus on risk/effect estimates such as HR and AHR together with measures of statistical precision (95% CI) to report and interpret the results, instead of P-values. The black and white approach inherent in statistical significance testing is strongly discouraged nowadays, being methodologically flawed though still largely used (see on this for instance PubMed PMIDs 18582619, 33865412, 28938712, 27209009, 27272951, 28938715, 29650628, 30230362, 30894741, 34453632 and the ASA statement doi.org/10.1080/00031305.2016.1154108). The authors could dismiss all P-values through the manuscript, since they tend always to be interpreted according to the conventional cutpoints of 0.05/0.001. They should instead focus on the size of the effects, dose-response or trend analysis, and stability of the estimates as assessed through the CI width (not just inclusion in it of the unit!). This may have major effect also in the interpretation of the results reported in the abstract: for instance, the authors claim that mortality in W2 – 23.7% - was (‘significantly’) higher than in W1 (21.3%) and W3 (15.8), but actually the absolute difference between W2 and W1 was small, and much smaller than the one between W2 and W3, and W1 and W3. In addition, again in the abstract the authors claim that there was no difference in mortality rates in patients >75 years, but they provide no figures for such a statement, that I suspect was made based on statistical significance testing/P-value cutpoints.

- In the Abstract, please provide the list of relevant confounders that were adjusted for in the analysis, and that allowed to assess W2 and W3 mortality as being ‘significantly lower’ than in W1.

- I see a risk of overadjustment in the large number of variables used to adjust mortality rates in the multivariable model comparing period-specific (W1-W2-W3) mortality. May the authors try also ‘intermediate’ or ‘progressive’ models in terms of adjustment factors, possibly adding the most relevant of them one by one? In addition and more importantly, the authors should carry out stratified analysis, for instance in specific subgroups with reference to sex, age groups, co-morbidities, clinical conditions at hospital admission.. Stratified analyses are always more powerful and informative than multivariable ‘adjusted’ analyses, allowing to identify which subgroups (of hospitalized patients in the present case) experienced different outcomes, independently from the fact that changes in specific subgroups was able to influence the overall estimates.

- The authors should elaborate much more on the potential causes of the decrease over time in the case-fatality rates of hospitalized patients. May also viral variants have played a role?

- The authors should better clarify if they use mortality rate and case-fatality rate is synonymous in the manuscript.

- Tables and Figures: the authors should report mortality rates of hospitalized patients by age groups and sex in the three waves, possibly by plotting them in figures, and also stratifying if possible by comorbidities and other variables such as critical COVID-19 at admission. This in order to clarify which factor(s) drove the mortality decrease over time, including a possible effect of COVID-19 therapy (as correctly hypothesized by the authors in the Discussion) and of vaccination in W3.

- In the Introduction, I strongly suggest the authors to assess more in-depth the literature, focusing on studies that in Italy and outside this country assessed mortality/case-fatality rate of COVID-19, inside and outside hospitals.

- May the authors take into consideration vaccination status in their analyses about changes over time of COVID-19 mortality in hospitalized patients?

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

11 Mar 2022

Replies to the Reviewers

Comments from Reviewer 1 :

We thanks the reviewer for his/her time and suggestions.

Comment #1: The one limitation that isn’t stated is that any choice of “waves” to divide the pandemic into is somewhat arbitrary. There are more mathematical ways to determine such waves rather than simply by inspection, as proposed in this paper “COVID-19 second wave mortality in Europe and the United States” (https://pubmed.ncbi.nlm.nih.gov/33810707/). It could probably be cited regardless as its subject matter is very similar to yours (reduction of mortality in subsequent Covid waves).

Replay:We fully agree with the Reviewer’s comment, and accordingly we have added among limitations of our study (end of the discussion section the following sentence: ‘the criteria that we have adopted for defining waves were based simply on the inspection of the shape of the Italian COVID-19 epidemic trajectory, but should be noticed that mathematical methods in time series analysis have been recently proposed to partition time series in order to better define waves and CFR during waves [Chaos. 2021;31(3):031105.].

Comment #2: There are some truly minor typos: co-morbidities->comorbidities;

up-dated->updated; re-organisation->reorganisation

Replay: We have corrected the typos in the text.

Comments from Reviewer 2 :

General comment This is an interesting study that analyzes relevant issues in the history of SARS-CoV-2 infection and COVID-19 disease in a specific and highly relevant Italian hospital. It assessed if and how COVID-19 mortality differed between three consecutive periods in which the knowledge and consequently the management of the disease improved considerably. However, I see a few major methodological issues and other points in the manuscript that need to be addressed in my opinion.

Replay: We thanks the reviewer for his/her time and constructive feed-back.

Comment #1: I strongly recommend the authors to dismiss statistical significance/null hypothesis testing in the entire study, including data analysis and interpretation. They should instead focus on risk/effect estimates such as HR and AHR together with measures of statistical precision (95% CI) to report and interpret the results, instead of P-values. The black and white approach inherent in statistical significance testing is strongly discouraged nowadays, being methodologically flawed though still largely used (see on this for instance PubMed PMIDs 18582619, 33865412, 28938712, 27209009, 27272951, 28938715, 29650628, 30230362, 30894741, 34453632 and the ASA statement doi.org/10.1080/00031305.2016.1154108). The authors could dismiss all P-values through the manuscript, since they tend always to be interpreted according to the conventional cutpoints of 0.05/0.001. They should instead focus on the size of the effects, dose-response or trend analysis, and stability of the estimates as assessed through the CI width (not just inclusion in it of the unit!). This may have major effect also in the interpretation of the results reported in the abstract: for instance, the authors claim that mortality in W2 – 23.7% - was (‘significantly’) higher than in W1 (21.3%) and W3 (15.8), but actually the absolute difference between W2 and W1 was small, and much smaller than the one between W2 and W3, and W1 and W3. In addition, again in the abstract the authors claim that there was no difference in mortality rates in patients >75 years, but they provide no figures for such a statement, that I suspect was made based on statistical significance testing/P-value cutpoints.

Replay: We thank the reviewer for the observation and we feel to agree with his/her comments. Accordingly, we have modified the text by presenting and interpreting the significance of our data only using the risk/effect estimates and confidence intervals.

Comment #2: In the Abstract, please provide the list of relevant confounders that were adjusted for in the analysis, and that allowed to assess W2 and W3 mortality as being ‘significantly lower’ than in W1.

Replay: We have added in the abstract the list of all of the relevant confounders we used to adjust our Cox models.

Comment #3: I see a risk of overadjustment in the large number of variables used to adjust mortality rates in the multivariable model comparing period-specific (W1-W2-W3) mortality. May the authors try also ‘intermediate’ or ‘progressive’ models in terms of adjustment factors, possibly adding the most relevant of them one by one?

Replay: We thank the reviewer for his/her suggestion. Attached below you can find the results of a model we have built by introducing one by one variable to an initial model in which we included only age, sex and obesity which could be considered as truly confounders in the casual path of the study. In subsequent models we have progressively introduced variables that could be considered as mediator of the effect, to reach a final model as that presented in the original version of our manuscript. It should be noticed that the estimates obtained for the different waves are consistent in each model with that provided by the final model which was previously presented, so we have preferred to maintain the latter also in view of the fact that all examined variables are commonly evaluated in clinical practice when evaluating the clinical conditions and prognosis of COVID-19 patients. We have not included in the revised version of our manuscript the results of the new analysis, but if the Editor/Reviewer considers it relevant for our data interpretation we could add it as supplementary material.

AHR AHR Lower bound (95%) AHR Upper bound (95%)

Age (for 10 year more) 1,952 1,801 2,116

Wave 2 vs. 1 0,748 0,594 0,942

Wave 3 vs. 1 0,543 0,415 0,710

Male vs. female 1,659 1,345 2,046

Obesity yes vs. no 1,758 1,411 2,191

AHR AHR Lower bound (95%) AHR Upper bound (95%)

Age (for 10 year more) 2,156 1,975 2,354

Wave 2 vs. 1 0,736 0,584 0,928

Wave 3 vs. 1 0,562 0,429 0,736

Male vs. female 1,360 1,101 1,679

Obesity yes vs. no 1,641 1,317 2,046

Disease severe vs. mild/moderate 2,119 1,652 2,717

Disease critical vs. mild/moderate 5,019 3,915 6,434

AHR AHR Lower bound (95%) AHR Upper bound (95%)

Age (for 10 year more) 2,184 1,997 2,388

Wave 2 vs. 1 0,747 0,589 0,946

Wave 3 vs. 1 0,583 0,444 0,767

Male vs. female 1,357 1,096 1,681

Obesity yes vs. no 1,656 1,325 2,070

Disease severe vs. mild/moderate 2,196 1,705 2,828

Disease critical vs. mild/moderate 5,029 3,899 6,486

Time from symptoms onset (x 1 day more) 0,995 0,979 1,011

AHR AHR Lower bound (95%) AHR Upper bound (95%)

Age (for 10 year more) 2,183 1,993 2,391

Wave 2 vs. 1 0,747 0,589 0,946

Wave 3 vs. 1 0,583 0,444 0,767

Male vs. female 1,357 1,095 1,682

Obesity yes vs. no 1,656 1,323 2,072

Disease severe vs. mild/moderate 2,196 1,705 2,829

Disease critical vs. mild/moderate 5,030 3,898 6,491

Time from symptoms onset (x 1 day more) 0,995 0,979 1,011

Num. comorb. ≥ 3 vs. < 3 1,004 0,788 1,280

Comment #4: In addition and more importantly, the authors should carry out stratified analysis, for instance in specific subgroups with reference to sex, age groups, co-morbidities, clinical conditions at hospital admission. Stratified analyses are always more powerful and informative than multivariable ‘adjusted’ analyses, allowing to identify which subgroups (of hospitalized patients in the present case) experienced different outcomes, independently from the fact that changes in specific subgroups was able to influence the overall estimates.

Replay: We thank the reviewer for the suggested data analysis. We have provided an additional Figure (S2) in which we have plotted the estimated survival probability of specific patients’ subgroups according to the period of their admission (W1, W2, and W3). The subgroups were defined by multiple clinically relevant variables including: sex, age (>75 vs ≤75 years), disease severity (critical vs non-critical) and comorbidity burden (≥3 comorbidities vs <3 comorbidities). Findings were described in the Results section as follows: ‘Results of survival analysis according to the period of admission of subgroups of patients stratified by multiple characteristics are presented in the Supplementary Figure 2. Overall, an improvement in survival was observed in subgroups of patients aged ≤75 years, which was more evident for males with critical disease or with multiple comorbidities. Conversely, there was no significant change in survival in all groups of patients aged> 75 years over the three epidemic waves.

Comment #5: The authors should elaborate much more on the potential causes of the decrease over time in the case-fatality rates of hospitalized patients. May also viral variants have played a role?

Replay: We fully agree with the Reviewer’s comment. At this regard, we have added the following statement in the the discussion section: ‘It is unlikely that changes in SARS-CoV-2 virulence may have played a role in the over time decrease of mortality rate we observed among our cohort of hospitalized patients. In fact, over W2 and W3 an increase in the circulation of SARS-CoV-2 variant alpha (B.1.1.7) has been observed in Italy [Virol J. 2021 Aug 14;18(1):168.], and some studies in UK have shown that this variant is associated with increased risk of critical care need and mortality compared to non-B.1.1.7 viral strains [Lancet Infect Dis. 2021 Nov;21(11):1518-1528]. We have also commented further on this issue as follows: ‘On the other hand, as the pandemic progresses, it is important continuing to monitor case fatality rate associated with COVID-19 and the impact that emerging viral variants and vaccination coverage have on the risk of hospitalization and COVID-19 death’.

Comment #6: The authors should better clarify if they use mortality rate and case-fatality rate is synonymous in the manuscript.

Replay: In our study conducted on a well defined population of patients hospitalized with COVID-19 the terms ‘mortality rate’ and ‘case-fatality rate’ can be considered synonymous.

Comment #7: Tables and Figures: the authors should report mortality rates of hospitalized patients by age groups and sex in the three waves, possibly by plotting them in figures, and also stratifying if possible by comorbidities and other variables such as critical COVID-19 at admission. This in order to clarify which factor(s) drove the mortality decrease over time, including a possible effect of COVID-19 therapy (as correctly hypothesized by the authors in the Discussion) and of vaccination in W3

Replay: We thank the Reviewer for his/her suggestion. a) We have provided a new Figure (S2) with the survival analysis for different subgroups stratified by multiple clinically relevant variables (see also response to comment 4). b) We do agree that the effect of different treatments could have impacted toward an improvement of outcome especially for those patients with a critical disease in which an improved knowledge of the disease’s natural history was maturated overtime. Nevertheless, the present study does not allow to draw any inference regarding the effect of different treatments apart from the modification of the standard of care from the 1st wave (oxygen supply plus remdesivir) to that of the 2nd and 3rd waves (oxygen supply plus remdesivir plus dexamethasone plus enoxaparin).

As regards the possible role of vaccination see our response to comment 9.

- Comment #8: In the Introduction, I strongly suggest the authors to assess more in-depth the literature, focusing on studies that in Italy and outside this country assessed mortality/case-fatality rate of COVID-19, inside and outside hospitals.

Replay: This is a correct observation and accordingly we have focused more on available data regarding overall CFR of COVID-19 reported in Italy and in other countries hit by the pandemic. We have added the following paragraph in the Introduction section:‘Initial analyses performed in April 2020 of data from 9 countries (China, France, Germany, Italy, the Netherlands, South Korea, Spain, Switzerland and Unites States) showed wide variation of the overall case fatality rate (CFR): from 0.7% in Germany to 9.3% in Italy, with two-thirds of the CRF variability being explained by differences in the age distribution of the cases [Ann Intern Med. 2020 Nov 3;173(9):714-720.]. In Italy, the CFR increased to 14% by June 2020, mostly attributable to increasing age-specific case-fatality [PLoS One. 2020 Sep 10;15(9):e0238904].’

- Comment #9: May the authors take into consideration vaccination status in their analyses about changes over time of COVID-19 mortality in hospitalized patients?

Replay: We might exclude a contributing role of COVID-19 vaccination in the decreased mortality rate of hospitalized COVID-19 patients we observed during W3. In fact, although W3 overalapped with the early phase of the Italian vaccination campaign, no patients in our cohort were fully vaccinated against COVID-19 and only two had received one dose of vaccine. We have added this statement in the Conclusion section.

Submitted filename: risposte Editor and reviewers comments R1 4-3-22.docx

Click here for additional data file.

30 Mar 2022

Mortality rates among COVID-19 patients hospitalised during the first three waves of the epidemic in Milan, Italy: a prospective observational study

PONE-D-22-01769R1

Dear Dr. Giacomelli,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Chiara Lazzeri

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

1 Apr 2022

PONE-D-22-01769R1

Mortality rates among COVID-19 patients hospitalised during the first three waves of the epidemic in Milan, Italy: a prospective observational study

Dear Dr. Giacomelli:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Chiara Lazzeri

Academic Editor

PLOS ONE