COVID-19: Spatial analysis of hospital case-fatality rate in France

Marc Souris; Jean-Paul Gonzalez

doi:10.1371/journal.pone.0243606

. 2020 Dec 15;15(12):e0243606. doi: 10.1371/journal.pone.0243606

COVID-19: Spatial analysis of hospital case-fatality rate in France

Marc Souris ^1,^*, Jean-Paul Gonzalez ^2,^3,⁴

Editor: Muhammad Adrish⁵

PMCID: PMC7737987 PMID: 33320895

Abstract

When the population risk factors and reporting systems are similar, the assessment of the case-fatality (or lethality) rate (ratio of cases to deaths) represents a perfect tool for analyzing, understanding and improving the overall efficiency of the health system. The objective of this article is to estimate the influence of the hospital care system on lethality in metropolitan France during the inception of the COVID-19 epidemic, by analyzing the spatial variability of the hospital case-fatality rate (CFR) between French districts. In theory, the hospital age-standardized CFR should not display significant differences between districts, since hospital lethality depends on the virulence of the pathogen (the SARS-CoV-2 virus), the vulnerability of the population (mainly age-related), the healthcare system quality, and cases and deaths definition and the recording accuracy. We analyzed hospital data on COVID-19 hospitalizations, severity (admission to intensive care units for reanimation or endotracheal intubation) and mortality, from March 19 to May 8 corresponding to the first French lockdown. All rates were age-standardized to eliminate differences in districts age structure. The results show that the higher case-fatality rates observed by districts are mostly related to the level of morbidity. Time analysis shows that the case-fatality rate has decreased over time, globally and in almost all districts, showing an improvement in the management of severe patients during the epidemic. In conclusion, it appears that during the first critical phase of COVID-19 ramping epidemic in metropolitan France, the higher case-fatality rates were generally related to the higher level of hospitalization, then potentially related to the overload of healthcare system. Also, low hospitalization with high case-fatality rates were mostly found in districts with low population density, and could due to some limitation of the local healthcare access. However, the magnitude of this increase of case-fatality rate represents less than 10 per cent of the average case-fatality rate, and this variation is small compared to much greater variation across countries reported in the literature.

Introduction

Since the beginning of the epidemic, the lethality rate, or case-fatality rate (CFR) of COVID-19 and the differences between countries have been the subject of many questions about national pandemic response policies and patient treatment. Most studies on the lethality of COVID-19 seek to estimate the true lethality of the disease, an issue that has been addressed since the beginning of the epidemic [1–4].

The case-fatality rate is the ratio between the number of deaths due to the disease and the number of closed cases (i.e. recovered or dead). It is estimated by the healthcare system based on the reporting of these two values. The case-fatality rate should not be confused with the mortality rate, which is the ratio of the number of deaths to the total population, or also with the morbidity rate, which is the ratio of the number of cases to the total population. Mortality and morbidity rates depend on the extent of disease in a population, unlike case-fatality rates [5].

The case-fatality rate of a disease in a population is an index of severity of the disease in that population, and of the capacity of the healthcare system to reduce mortality. In principle, this allows to compare the effectiveness of healthcare systems across regions or countries.

The aim of this article is to analyze the effectiveness of the healthcare system in France in the context of the COVID-19 epidemic. This study also discusses the differences and the comparison on case-fatality rates published by the international agency by country (May 2020).

Lethality depends on the virulence of the virus but, unlike morbidity, it does not depend on its contagiousness. Virulence comes from the reproductive capacity of the virus in the cell, its capacity for cellular degradation, and its ability to induce or not an innate or specific immune response. Virulence is of purely biological origin and once the virus has entered the target cell where it will cause its pathogenic effect does no longer depends on environmental conditions outside the host. Virulence is independent of the host population but may change over time and space if there is a risk of natural mutation/selection of the pathogen [6–8]. Contagiousness characterizes the biological capacity of the virus to reach the target cell system of its host, and the ability to be transmitted from one individual to another. The efficiency of transmission depends largely on environmental conditions (e.g., climate, urbanization, population density, mobility), which can vary greatly from one country to another.

In addition to the virulence of the virus, the case-fatality rate depends on viral load, on biological risk factors and on population vulnerability (age structure, genetic factors, prevalence of co-morbidities, healthcare accessibility, etc.) as well as other factors related to the health system (equipment, capacity, staff, management, care of patients, effectiveness of therapies, patient management in a critical phase of the disease), and factors related to the detection and registration system for cases and deaths (clinical cases definition, detection, surveillance systems, case and death reporting). The evaluation of the case-fatality rate normally requires the detection and counting of all infected persons, irrespective of their level of symptoms (i.e. disease severity).

When the population risk factors and reporting systems are identical, case-fatality rate evaluation represents an excellent tool for analyzing, understanding and improving the overall performance of the health system, particularly at the level of hospital units. Studying the magnitude of differences in case-fatality rates between units also makes it possible to assess the impact of the quality of the health system on case-fatality.

There are large differences in the case-fatality rates of COVID-19 published by country (Table 1) or calculated directly from WHO data. These rates vary considerably, from less than 0.02 (Thailand, Australia, Chile) to more than 0.15 (France, Belgium, UK), with a mean at 0.04 and a standard deviation of 0.045 (WHO, May 8, 2020, Fig 1).

Table 1. Case-fatality rates and characteristics of countrywide population vulnerability to COVID-19.

Country	CFR (%)^*	Population > 65-year-old (%)^**	Hospital bed number per 100,000 people^***
France	19.00	20.02	621
Belgium	16.42	18.80	623
UK	14.94	18.40	273
Italy	13.84	22.75	331
Netherlands	12.59	19.20	466
Sweden	12.29	20.10	254
Spain	11.73	19.38	297
Iran	6.31	6.18	150
USA	5.40	15.80	290
Greece	5.52	21.65	424
China	5.50	10.92	420
Switzerland	5.01	18.62	470
Germany	4.28	21.46	823
Portugal	4.15	21.95	332
Austria	3.88	19.02	760
Tanzania	3.75	2.60	70
Japan	3.56	27.58	1,340
India	3.36	6.18	70
Czech Republic	3.28	19.42	645
Turkey	2.72	8.48	270
Norway	2.62	17.05	390
Uruguay	2.53	14.81	280
South Korea	2.36	14.42	1150
Thailand	1.83	11.90	210
Slovakia	1.74	15.63	579
Australia	1.41	15.65	380
Chile	1.21	11.53	220

Open in a new tab

*Lethality rate for Covid-19 reported among selected countries from WHO, as for May 8, 2020 (https://covid19.who.int/)

**population percentage over 65-year-old by World Bank, 2016

***Hospital beds per 100,000 people by World Bank, 2013, Eurostat, 2014.

In Europe, the characteristics of populations (in terms of risk factor for COVID-19) and health systems are comparable, but the definition, detection and reporting of cases and causes of death can differ greatly from one country to another. Some countries conducted significantly more detection tests and hospitalizations than others (Table 2), resulting in differences in the protocols for patient management. The rate of testing performed (policy) and mortality rates (reporting) vary mainly according to the geographical extent of the epidemic within each country.

Table 2. SARS-Cov-2 detection, COVID-19 morbidity and mortality rates by selected country.

Country	rtPCR Test number per 1,000*	Morbidity rate (cases per million)	Mortality rate (death per million)
Belgium	44	4488	735
Spain	41	5494	558
Italy	39	3570	495
Germany	33	2022	88
Australia	29	271	4
United States	25	3906	232
United Kingdom	22	3045	451
France	21	2678	398
Turkey	15	1586	43
Netherlands	14	2438	309
South Korea	12	211	5
Iran	6	1246	78
Japan	1.5	122	4.5
India	1	41	1.4
Indonesia	0.5	48	3.5

Open in a new tab

Source: Worldometer, May 8, 2020.

The virulence of the COVID-19 pathogen (SARS-CoV-2 virus) is assumed to be identical in all countries. In order to compare case-fatality rates across regions or countries (and thus analyze the effectiveness of healthcare systems), it is necessary, when calculating rates, to standardize population-related risk factors and to use the same definitions and enumeration methods to record cases and deaths. This is not the case for the current pandemic and discrepancies exist among the country systems.

The principal objective of this article is not to estimate the actual lethality of COVID-19 in France based on the rates published by the health authorities, but to estimate the influence of the healthcare system on lethality by analyzing the spatial variability of the hospital case-fatality rate (confirmed hospitalized cases and hospital deaths) in metropolitan France between districts (i.e. French départements). This analysis, limited to metropolitan France, is possible while it remains within the framework of the same system for defining and counting cases and deaths. We thus assume that this system of definition and enumeration was identical throughout France during the period (19 March to 8 May) corresponding to the first wave (inception) of the COVID-19 epidemic in France. Therefore the study focuses on the extent of spatial differences in the case-fatality rate in metropolitan France, and enables to highlight the relative differences between districts, as well as to analyze the causes independently of the system of definition and enumeration of cases and deaths, and also independently of the main biological risk factor of severity (age) after standardization on this factor.

The other objective is to analyze the case-fatality rate observed in metropolitan France with the one calculated for other countries: the variability of the case-fatality rate between French districts (mainly due to patient management in the acute epidemic phase) allow us to estimate whether this observed variability can explain the significant differences in case-fatality rates observed between France and other countries.

Data and methods

Data

This study is based on daily hospitalization and death declaration data by district in France and is accessible on the “Santé Publique France” website. (www.data.gouv.fr/fr/datasets/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19) from March 19 to May 8, 2020, corresponding to 50 days lockdown (i.e. quarantine) and the spread of the COVID-19 epidemic in France (Fig 2). We used data on the distribution of hospitalized COVID-19 cases according to age group (10-year age group) (Santé Publique France). We also obtain demographic data by districts (source: population by age, INSEE, 2020, www.insee.fr), healthcare system (number of hospital beds) and some health status rates (hypertension, coronary disease, acute respiratory disease, diabetes) by districts (source: Ecosanté France, www.ecosante.fr, 2016),

Fig 2 — COVID-19 timeline of hospitalizations (left) and death (right), March, 19 to May, 8 2020, France.

This analysis was carried out on the 96 districts of metropolitan France (Fig 3), while the French overseas districts and territories were excluded from the analysis for reasons of spatial analysis and mapping. The data were integrated into a geographic information system (SavGIS, ww.savgis.org) for analysis and mapping.

Focusing only on confirmed hospitalized cases and deaths in hospital, the case-fatality rates calculated in this article for France refer to the hospital case-fatality rates and do not represent the absolute national case-fatality that should include all positive cases and deaths related to COVID-19. Given the estimated high number of non-severe and asymptomatic forms (which a fortiori do not cause deaths)—it is estimated that only 2.6% of infected persons were hospitalized [9]—this overall lethality is necessarily much lower than hospital lethality, but it will be accurately calculated only at the end of the epidemic when the total number of positive cases (i.e. seroprevalence survey) will be available and the total number of deaths outside hospital due to COVID-19 will be accurately assessed.

All identified and hospitalized cases were tested positive (by rtPCR). All deaths counted were COVID-19 associated. As of May 8, 2020, not all hospitalized cases are closed since the epidemic is still ongoing: deaths counted at the beginning of the study period correspond to cases hospitalized but were not included in the study, and cases counted at the end of the period were not closed and no deaths from these cases were included in the study.

The hospital healthcare system in France offer a mean of 621 Hospital beds per 100,000 people. The hospital healthcare system is fairly evenly distributed throughout the metropolitan territory, as shown in Table 3 for regions (www.ecosante.fr).

Table 3. Hospital bed number per 100,000 people, per regions in France.

Region	Name	Hospital bed number per 100,000 people
France	Whole country	621
11	Île de France	547
24	Centre-Val de Loire	634
27	Bourgogne-Franche-Comté	711
28	Normandie	621
32	Hauts-de-France	634
44	Grand Est	642
52	Pays de la Loire	569
53	Bretagne	651
75	Nouvelle Aquitaine	686
76	Occitanie	691
84	Auvergne-Rhône-Alpes	659
93	Provence-Alpes-Côte d’Azur	778
94	Corse	678

Open in a new tab

Methods

Age standardization

Since age is a major risk factor for death for COVID-19 patients, to compare hospitalization rates, mortality rates, or case-fatality rates between districts it’s important to eliminates the differences between districts due to differences in age structure of the population of the districts. We performed an indirect standardization [10] by calculating the rate (hospitalization rate and hospital lethality rate) per ten-years age group for the whole of France. We also estimated lethality rate for the whole of France excluding areas that have been under hospital stress (Ile-de-France and Grand-Est), in order to evaluate hospital case-fatality rates per age group independently of possible excess mortality due to the saturation of healthcare systems in certain districts. Classically, E_i (expected number of deaths in district i) is the sum over the different age groups of the district's population in the age group multiplied by the hospital case-fatality rate of the age group:

E_{i} = \sum_{a g e c l a s s a} P_{i, a} \times T_{a}

where P_i,a is the population of the age group “a” of the district i and T_a is the hospital case-fatality rate of the age group calculated over the whole country.

Standardized case-fatality rate

The Standardized Lethality Ratio (SLR) is equal to the ratio between the observed death toll Oi and the expected death toll Ei. The standardized hospital case-fatality rate LS_i for each district is equal to the average rate for France multiplied by the district's SLR. The significance of the SLR (H0: SLR = 1, no statistically significant over—or under-fatality compared to the expected value) is evaluated for each district by a Breslow & Day test [11, 12] in order to account for the statistical significance of the differences in these ratios between districts. In order to take into account the problem of multiple testing [13] when the null hypothesis concerns the whole territory (“is there a district for which the H0 hypothesis is rejected?”), the risk of error was set at 0.0005 in order to apply a Bonferonni correction (which is very conservative when data is spatially autocorrelated). When the study is conducted for a single district (“Is the SLR for this district different from 1?”), the risk of error is set at 0.05. All tests are bilateral.

Besides age, another risk factors for severity and mortality have been observed for COVID-19, including hypertension, cardiovascular disease history, diabetes, obesity. Even if these risk factors are often related to age, to detect another risk factor for lethality (as healthcare saturation) using districts comparison, it would be useful to standardize on these factors to. Unfortunately, apart for age, we do not have the necessary data to evaluate the global case-fatality rates due to COVID-19 for these risk factors or combination of factors. To overcome this difficulty, we analyzed relationships between CFR and health risk factors rates (diabetes, coronary disease, hypertension, respiratory syndrome), but these correlations should be used with caution due to possible ecological fallacy (they should be done at individual level, but these data are not available).

Cartography

The values of morbidity rates, mortality rates, non-standardized case-fatality rates, SLRs and their significance were mapped by districts. In order to map case-fatality rates independently of the rate calculation variability, due to differences in hospitalization numbers between districts, rates adjusted with empirical Bayesian estimator (EBE) were calculated [14]. All statistical and spatial analysis and mapping were carried out using the SavGIS geographic information system v. 9.15 (www.savgis.org).

Study of statistical correlations

In order to estimate whether possible congestion in the healthcare system influenced hospital case-fatality rate, we studied the correlations between hospitalization and hospital lethality by district over the entire period. Inpatient severity was estimated using the ICU rate (number of patients admitted in intensive care /number of inpatients). We analyzed also relationship between standardized CFR and healthcare system capacity (bed number per people).

Results

For the whole of France (data from 19 March to 8 May 2020), the hospital case-fatality rate was 0.174. The average hospital case-fatality rates by age group, used in indirect standardization, was calculated from French Public Health data by age class published on 8 May 2020 (Table 4).

Table 4. Hospitalization, mortality and case-fatality rates by age group in metropolitan France (2020).

Age class	Population (%)	Hospitalization
		Hospitalized		Death		CFR
		Total	%	Total	%	Whole country	Without #11 & 44 regions	#11 & 44 regions only
0–9	11.43	617	0.65	2	0.01	0.0032	0	0.0049
10–19	12.27	463	0.49	3	0.02	0.0065	0	0.0125
20–29	11.15	2253	2.4	19	0.11	0.0084	0.0051	0.0121
30–39	12.36	4325	4.5	72	0.44	0.0166	0.0088	0.0229
40–49	12.80	7001	7.4	203	1.24	0.0290	0.0183	0.0372
50–59	13.10	12130	12.9	747	4.56	0.0616	0.0435	0.0751
60–69	12.01	16115	17.1	1934	11.81	0.1200	0.0988	0.1361
70–79	8.57	18812	20.0	3723	22.73	0.1979	0.1666	0.2244
80–89	4.91	21625	22.9	6192	37.80	0.2863	0.2688	0.3033
90+	1.38	10897	11.5	3485	21.28	0.3198	0.3047	0.3346
All ages	100	94238	100	16380	100	0.1738	0.1597	0.1858

Open in a new tab

(Region 11 = Ile-de-France; region 14 = Grand-Est).

The epidemic in France (data from 19 March to 8 May 2020) presents a very uneven spatial distribution of hospitalizations. It has spread throughout the country at low level, with a mean hospitalization rate of 1.27 per 1,000 (median at 0.88) and a standard deviation of 1.03, except in some regions where transmission has been much more active, such as the Ile-de-France region (mean hospitalization rate 3.01 per 1,000), or the Grand-Est region (mean hospitalization rate 3.2 per 1,000). The mean hospitalization rate varies from 0.19 per 1,000 (Tarn-et-Garonne) to 4.75 (Territoire de Belfort) (Fig 4).

Fig 4 — Hospitalization rate (left) and hospital mortality rate (per 1000 inhabitants) (right) for COVID-19 in metropolitan France for the period March 19 to May 8, 2020.

The hospital mortality rate (not age-standardized) has the same spatial distribution. It varies from 0.01 per 1,000 (Tarn-et-Garonne) to 1.13 per 1,000 (Territoire de Belfort), with a mean of 0.21 (median 0.12) and a standard deviation of 0.21.

Such relative differences in the circulation of the disease reported by districts in French metropolitan area clearly appears in the cartogram displayed below (Fig 5).

Fig 5 — The geometry of the districts in France is shown in the lower left corner.

The observed gross hospital case-fatality rate calculated by district showed a mean of 0.153 with a standard deviation of 0.045. It varies between 0.046 (Ariège) and 0.258 (Vosges). Without Ile-de-France and Grand-Est regions, the mean gross case-fatality rate per district is 0.149 with a standard deviation of 0.048 (Fig 6).

Fig 6 — Case-fatality rate (left) and case-fatality rate adjusted by empirical Bayesian estimator (right) for the COVID-19 epidemic in France from 19 March to 8 May 2020.

Case-fatality rates have evolved over the period of the study. Here we represent non-standardized rates calculated for three periods: March 19 to 31; April 1 to 12; April 13 to 25. Rates are calculated with a 12-day time lag between cases and deaths (Fig 7).

The standardized case-fatality rate per district is between 0.044 (Ariège, SLR = 0.25) and 0.255 (Vosges, SLR = 1.48), with the mean at 0.151 and the median at 0.153. If the Ile-de-France and Grand-Est regions are excluded from the calculation of age-specific case-fatality rates, the CFR is between 0.044 and 0.251, with the mean at 0.144 and a median at 0.147 (SLR between 0.28 and 1.67, mean at 0.99 and median at 1) (Fig 8). In the following, we will consider only the standardized case-fatality rates calculated with age-specific case-fatality rates that do not take into account the Ile-de-France and Grand-Est regions.

The spatial distribution of standardized morbidity rate (hospitalized cases) shows a significant spatial autocorrelation (Moran index: 1.54, p-value < 10⁻⁶), and this is expected for an infectious disease [14, 15]. The case-fatality rate shows also significant spatial autocorrelation (Moran index: 0.29, p-value < 0.000007), and this is no expected. The analysis of the clusters clearly shows a clustering of high case-fatality rate values in regions of high morbidity (particularly the Grand-Est), and shows some cases of districts with high case-fatality rate values isolated in areas with low rates.

The Breslow & Day significance test for SLR shows districts where the SLR is statistically significantly different from 1, corresponding to districts with abnormally high (SLR > 1, red) or abnormally low (SLR < 1, green) standardized case-fatality rates. The individual significance threshold is set at 0.05, and for all districts at 0.0005 to account for multi-testing (Fig 9).

Fig 9 — Geographical units (districts) with SLR significance: P-value <0.05 (left), p-value < 0.0005 (right).

There is no correlation between hospital capacity (bed number per 100,000) and COVID-19 standardized case-fatality rate per district (Bravais-Pearson index r = -0.11).

At district level, there is no significant correlation (Bravais-Pearson index r = 0.14) between case-fatality rate and health risk factors rate (coronary disease, hypertension, acute respiratory disease, diabetes). Further investigation with patient data will be necessary to analyze the possible influence of individual health risk factors on SLR at district level

However, there is a significant correlation between the standardized hospitalization rate and the standardized case-fatality rate (Bravais-Pearson index r = 0.40) (Fig 10). The correlation increases (0.48) if we limit the calculation to districts whose SLR is significantly (p-value < 0.05) different from 1.

To illustrate the increase of case-fatality rate with hospitalization rate, Table 5 gives the mean of the standardized case-fatality rate over the districts according to their standardized hospitalization rate. The average case-fatality rate varies from 0.134 for districts with a low hospitalization rate (less than 0.5 per 1000, 31 districts) to 0.152 for all 96 districts. If only districts with high hospitalization rates (above 2 per 1000, 16 districts) are considered, the average case-fatality rate is 0.183. For the eight districts with the highest hospitalization rates (> 3 per 1,000), and considered as district under hospital stress, the average case-fatality rate is 0.191 (Meuse, Moselle, Seine-St-Denis, Hauts-de-Seine, Paris, Val-de-Marne, Haut-Rhin, Territoire de Belfort), while it is 0.149 for all other districts.

Table 5. Average standardized CFR as a function of hospitalization rate during the COVID-19 epidemic in France for the period of time between March 19 to May 8, 2020.

Hospitalization rate (per 1000)	Districts number	Standardized CFR (mean)	p-value of T-test with World mean of CFR
< 0.5	31	0.134	< 10⁻⁶
< 1	59	0.138	< 10⁻⁶
< 1.5	72	0.142	< 10⁻⁶
< 2	80	0.146	< 10⁻⁶
< 2.5	86	0.147	< 10⁻⁶
< 3	89	0.149	< 10⁻⁶
< 3.5	90	0.149	< 10⁻⁶
< 4	93	0.150	< 10⁻⁶
All districts	96	0.152	< 10⁻⁶

Open in a new tab

A T-test shows that the mean of standardized CFR in France is very significantly (p-value < 10⁻⁶) higher than the world CFR average, even for districts with low hospitalization rate.

The mapping of the hospitalization rate and the hospital mortality rate with the SLR shows the strong spatial correspondence of these values (Fig 11).

Fig 11 — SLR (symbols) and hospitalization rate (left) and fatality rate (right) during the COVID-19 epidemic in France for the period of time between March 19 to May 8, 2020.

A typology combining hospitalization rates and case-fatality rates is proposed: low rates (values below the mean by less than one standard deviation), high rates (values above the mean by more than one standard deviation), so as to represent five classes (low-low, low-high, high-low, high-high, other) (Fig 12). The "high hospitalization-high case-fatality rate" class is spatially clustered. The "low hospitalization rate-high lethality rate" class is dispersed, and the population density is significantly lower in that class than in all the other districts (p-value < 0.02).

We call “severity rate” the ratio between the rate of patients in intensive care unit (for reanimation or endotracheal intubation) and the rate of hospitalization. It gives in principle an indication of the severity of the patients in hospital. At district level, the hospitalization rate and this severity rate show a low negative correlation (r = -0.22), indicating a decrease in intensive care rate when the hospitalization rate is high. This trend may be due to the saturation of intensive care units. The relationship between hospitalization and severity could also be interpreted as a decrease in less severe hospitalizations in order to be able to manage more severe cases when the healthcare system is overloaded, which would result in an increase in lethality. Nevertheless, in both cases, there is no correlation between the severity rate and the case-fatality rate (r = -0.10), indicating that globally, the differences in use of resuscitative measures or endotracheal intubation does not impact the case-fatality rate. Finally, the severity rate does not have a spatial distribution corresponding to the spatial distribution of the hospitalization rate (Fig 13).

Discussion

The standardized case-fatality rates at district level in France (0.04 for Tarn-et-Garonne to 0.26 for the Vosges district) remain in a ratio of 0.3 to 1.6 compared with the national average (mean 0.14, calculated by excluding districts under hospital stress in order not to take account of possible overloading of the healthcare systems). The relationships between standardized morbidity rates and standardized case-fatality rates in France show a correlation between these two indices, with an increase of SLR with hospitalization rate as shown in Table 5. It is therefore very likely that high medical admission rates and healthcare workers stress over the period under consideration has increased the hospital case-fatality rate: for the 20 districts with the highest hospitalization rates (essentially located in the Grand-Est and Ile-de-France regions), the average case-fatality rate is 20% higher than the average for all districts, and 25% higher than the average for all other districts alone. It is therefore likely that pressure on hospitals have increased the national average case-fatality rate by district from 0.145 to 0.153. It can thus be estimated that 2,425 deaths (out of the 16,732 deaths due to COVID-19 in hospital in France from 19 March to 8 May 2020, i.e. 15% of the total number of deaths) are due to the overload of the health system in the Grand-Est and Ile-de-France regions.

As we see, after standardization on age, lethality health risk factors (i.e. co-morbidities) do not show, at the district level, any significant correlation with differences in hospital lethality rates. With regard to behavioral risk factors associated to hospital lethality (e.g. late recourse to hospital care), studies focusing on mortality (i.e. not lethality) show significant differences in excess mortality according to the geographical origin of birth of the deceased [16, 17]. The authors of these studies attribute these differences in mortality to differences in the spatial distribution (the foreign-born population is more present in densely populated urban areas, like Ile-de-France region), and not to a risk factors for lethality which could be specific to this population.

There are clearly two risk profiles:

the districts where a high rate of hospitalization is coupled with a high case-fatality rate. These districts correspond to the epicenters of the epidemic, are clustered, and include the densely populated areas of Ile-de-France.
the districts where a low rate of hospitalization is coupled with a high case-fatality rate. These districts are geographically dispersed, sparsely populated and mainly located in rural areas.

The first profile accounts for 60% of the patients (59,878 over 99,970). The case-fatality average for this profile is significantly higher (0.187) as compared with the other districts case-fatality average (0.138). The profile displays a 33% excess mortality, corresponding to 2,716 deaths (11,003 observed vs. 8,287 expected with the CFR 0.138). This excess mortality, corresponding to 16% of all deaths, is likely to be the result of increased lethality due to overloading of the healthcare system.

The second profile (more rural) accounts for only 12.5% of the patients (12,514 over 99,970), with an average case-fatality rate significantly high (0.178) corresponding to a 25% excess mortality (2,170 observed vs. 1,731 expected deaths), but of low magnitude. It is possible that this excess mortality is due to a lack of care (medical desert, poor hospital response), which would have led to an increase in hospital lethality.

Some districts in the south of France have both a very low rate of hospitalization and a very low case-fatality rate (Gironde, Dordogne, Gers, Pyrénées Orientales), as a result of a low circulation of the virus and an effective response of the health system. Another particular case, is the one of the Bouches-du-Rhône, which appears with a high hospitalization rate (1.96 per 1,000, for a national average of 1.27) and a low case-fatality rate (0.11 for the Bouches-du-Rhône for a national average per district of 0.15). This situation could be due to a medical policy of more active screening and earlier hospitalization of the infected in the district, particularly in Marseille metropolitan area. The high value of the hospitalization rate would then be more the result of a non-severe treatment of the infected than of a more active circulation of the virus in the region, as for the lower hospitalization rates in neighboring districts (0.9 for Var, 0.56 for Vaucluse, 0.52 for Gars, 0.86 for Alpes-de-Haute-Provence, 0.81 for Alpes-Maritimes, 0.68 for Hérault). However, patient severity profile (expressed as the ratio between the number of patients admitted to intensive care and the total number of patients hospitalized) is not particularly low in the Bouches-Du-Rhône (0.163 for a mean of 0.156 and a standard deviation of 0.06 for all the districts), and this hospitalization policy is therefore not proven to be the primary factor in the low observed case-fatality rate.

Time analysis shows that the case-fatality rate has decreased over time, globally and in almost all districts, showing an improvement in the management of severe patients during the epidemic. This decrease stabilized rapidly and no longer evolved, in a steady manner. However, this result is limited by the difficulty of calculating a case-fatality rate on aggregated data over a short period of time, with eventual death occurring several days after hospitalization. We chose to delay the calculation of the number of deaths by 12 days compared to the calculation of the number of hospitalizations, since 85% of deaths occur on an average of 13 days after hospitalization [9].

WHO data from 219 countries show a mean case-fatality rate of 0.041 (median 0.029, standard deviation 0.045) with values between 0.31 (Nicaragua) and 0.00 (for countries that did not report deaths until May 8, like Vietnam). Even if we consider only the average case-fatality rate calculated only for the French districts with the lowest hospitalization rates (thus not causing possible saturation of the healthcare system), the French CFR average is very significantly higher than the World average (and the rates of most European countries, such as Spain, 11.73, Greece, 5.52, Germany, 4.28, etc.) (Table 1). The internal variability within France appears too low to explain the large differences in case-fatality rate between countries. These differences cannot be either explained by dissimilarities in morbidity between countries, in lethality risk factors between countries, or inpatient treatment, taking into account the quality of the healthcare system in France (Table 1). One can concluded that the difference between the case-fatality rate calculated with France datasets and the case-fatality rates presented using international WHO data is highly probably the result of a disparity in the registration of cases and/or deaths and not due to the quality of healthcare system or treatment. These differences in the counting of cases and/or deaths may be due to the hospitalization and screening policy specific to each country as well as the ability or willingness to hospitalize more non-severe forms, to the differences in case definition, or to insufficient quality of the health system for detecting and reporting cases and deaths. Such finding is reinforced by studies on excess mortality (difference to the average mortality of previous years over the same period) showing that in most countries, excess mortality is not fully explained by deaths officially attributed to COVID-19, while France datasets present a slight excess of COVID-19 reported deaths [18].

Conclusion

This study shows that the higher case-fatality rates observed in France in some districts during the first wave of the COVID-19 epidemic (data from 19 March to 8 May 2020) are mostly linked to the level of morbidity in the district, and therefore likely to the congestion of the healthcare systems during the acute phases of the epidemic. When the hospitalization rate is low, high case-fatality rates concern rural districts with low population density and could be linked to healthcare access in these districts.

However, the increase in the standardized case-fatality rate due to exceptional situations during epidemic peaks represents less than 10% of the average case-fatality rate per district in France, and the hospital case-fatality rate without these districts would be reduced from 0.151 to 0.138. This increase cannot therefore explain the extent of the difference observed between the average case-fatality rate in France and the average of the rates reported for all countries by international organizations or information sites (WHO, Wordometer, etc.). These differences probably stem from the reporting of cases and deaths, which is uneven from one country to another, and not from the care or treatment of patients during hospital stress due to epidemic peaks.

Data Availability

The data underlying the results presented in the study are available from www.data.gouv.fr/fr/datasets/donnees-hospitalieres-relatives-a-lepidemie-de-covid-19. The authors of the present study had no special privileges in accessing these datasets which other interested researchers would not have.

Funding Statement

The authors received no specific funding for this work.

References

1.Morteza AK, Fakher R. Cross-Country Comparison of Case Fatality Rates of COVID-19/SARS-COV-2. Osong Public Health and Research Perspectives 2020; 11(2): 74–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Rinaldia G, Paradisiba M. An empirical estimate of the infection fatality rate of COVID-19 from the first Italian outbreak. Harvard Economics Department. [Google Scholar]
3.Spychalski P, Blazynska-Spychalska A, Kobiela J. Estimating case fatality rates of COVID-19. The Lancet, 10.1016/S1473-3099(20)30246-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Baud D, Qi X, Nielsen-Saines K, Musso D, Pomar L, Favre G. Real estimates of mortality following COVID-19 infection. Lancet Infect Dis. 2020. (published online March 12). 10.1016/S1473-3099(20)30195-X [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Porta M. A dictionary of Epidemiology. 5th Ed Oxford University Press, Oxford, 2008. [Google Scholar]
6.Collier L, Oxford J. Human Virology. Oxford University Press, 2016. [Google Scholar]
7.Saluzzo JF, Vidal P, Gonzalez JP. Les virus émergents. IRD Edition, 2004. [Google Scholar]
8.Hurraux JM, Nicolas JC, Agut H, Peigne-Lafeuille H. Traité de virologie humaine. Estem/AUF, 2003. [Google Scholar]
9.Salie H, Tran-Kiem C, Lefrancq N, Courtejoie N, Bosetti P, et all. Estimating the burden of SARS-CoV-2 in France. 2020. Science 13 May 2020. 10.1126/science.abc3517 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Bouyer J, Hémon D, Cordier S, Derriennic F, Stûcker I, Stengel B, et al. Epidémiologie, principes et méthodes quantitatives. Inserm, 1993. [Google Scholar]
11.Breslow NE and Day NE, Statistical methods in cancer research, vol. I & II. WHO, International Agency for Reseach on Cancer, 1987. [Google Scholar]
12.Samuels SJ, Beaumont JJ, Breslow NE (1991). Power and detectable risk of seven tests for SMR. Am J Epidemiology 1991; 133:1191–1197. [DOI] [PubMed] [Google Scholar]
13.Miller RG. Simultaneous Statistical Inference. Springer Verlag, 1996. [Google Scholar]
14.Souris M. Epidemiology and Geography. Principles, methods and tools of spatial analysis. Wiley-ISTE, London, 2019 [Google Scholar]
15.Pfeiffer D, Robinson T, Stevenson M, Stevens K, Rogers D, Clements A. Spatial Analysis in Epidemiology. Oxford University Press, 2008. [Google Scholar]
16.Allard T, Bayardin V, Mosny E. L’Île-de-France, région la plus touchée par le surcroît de mortalité pendant le confinement. Insee Analyses Île-de-France n° 118, Juin 2020. [Google Scholar]
17.Papon S, Robert-Bobée I. Une hausse des décès deux fois plus forte pour les personnes nées à l’étranger que pour celles nées en France en mars-avril 2020. Insee Focus No 198—Juillet 2020 [Google Scholar]
18.Le Temps. L’impact du Covid-19 sous-estimé dans de nombreux pays. Available online: https://labs.letemps.ch/interactive/2020/surmortalite-et-covid-19/

PLoS One. doi: 10.1371/journal.pone.0243606.r001

Decision Letter 0

Muhammad Adrish

30 Jul 2020

PONE-D-20-17129

COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France

PLOS ONE

Dear Dr. Souris,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

==============================

ACADEMIC EDITOR: I have received the comments of the reviewers on your manuscript. The specific comments of the reviewers are included below. Please provide point by point response in your revised manuscript.

==============================

Please submit your revised manuscript by due date. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

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: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Muhammad Adrish

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

2.We suggest you thoroughly copyedit your manuscript for language usage, spelling, and grammar. If you do not know anyone who can help you do this, you may wish to consider employing a professional scientific editing service.

Whilst you may use any professional scientific editing service of your choice, PLOS has partnered with both American Journal Experts (AJE) and Editage to provide discounted services to PLOS authors. Both organizations have experience helping authors meet PLOS guidelines and can provide language editing, translation, manuscript formatting, and figure formatting to ensure your manuscript meets our submission guidelines. To take advantage of our partnership with AJE, visit the AJE website (http://learn.aje.com/plos/) for a 15% discount off AJE services. To take advantage of our partnership with Editage, visit the Editage website (www.editage.com) and enter referral code PLOSEDIT for a 15% discount off Editage services. If the PLOS editorial team finds any language issues in text that either AJE or Editage has edited, the service provider will re-edit the text for free.

Upon resubmission, please provide the following:

The name of the colleague or the details of the professional service that edited your manuscript
A copy of your manuscript showing your changes by either highlighting them or using track changes (uploaded as a *supporting information* file)
A clean copy of the edited manuscript (uploaded as the new *manuscript* file)

3. Please amend either the abstract on the online submission form (via Edit Submission) or the abstract in the manuscript so that they are identical.

4.We note that [Figure(s) 3, 4, 5, 6, 7, 8, 9, 11, 12 and 13] in your submission contain [map/satellite] images which may be copyrighted. All PLOS content is published under the Creative Commons Attribution License (CC BY 4.0), which means that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution. For these reasons, we cannot publish previously copyrighted maps or satellite images created using proprietary data, such as Google software (Google Maps, Street View, and Earth). For more information, see our copyright guidelines: http://journals.plos.org/plosone/s/licenses-and-copyright.

We require you to either (1) present written permission from the copyright holder to publish these figures specifically under the CC BY 4.0 license, or (2) remove the figures from your submission:

1. You may seek permission from the original copyright holder of Figure(s) [3, 4, 5, 6, 7, 8, 9, 11, 12 and 13] to publish the content specifically under the CC BY 4.0 license.

We recommend that you contact the original copyright holder with the Content Permission Form (http://journals.plos.org/plosone/s/file?id=7c09/content-permission-form.pdf) and the following text:

“I request permission for the open-access journal PLOS ONE to publish XXX under the Creative Commons Attribution License (CCAL) CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). Please be aware that this license allows unrestricted use and distribution, even commercially, by third parties. Please reply and provide explicit written permission to publish XXX under a CC BY license and complete the attached form.”

Please upload the completed Content Permission Form or other proof of granted permissions as an "Other" file with your submission.

In the figure caption of the copyrighted figure, please include the following text: “Reprinted from [ref] under a CC BY license, with permission from [name of publisher], original copyright [original copyright year].”

2. If you are unable to obtain permission from the original copyright holder to publish these figures under the CC BY 4.0 license or if the copyright holder’s requirements are incompatible with the CC BY 4.0 license, please either i) remove the figure or ii) supply a replacement figure that complies with the CC BY 4.0 license. Please check copyright information on all replacement figures and update the figure caption with source information. If applicable, please specify in the figure caption text when a figure is similar but not identical to the original image and is therefore for illustrative purposes only.

The following resources for replacing copyrighted map figures may be helpful:

USGS National Map Viewer (public domain): http://viewer.nationalmap.gov/viewer/

The Gateway to Astronaut Photography of Earth (public domain): http://eol.jsc.nasa.gov/sseop/clickmap/

Maps at the CIA (public domain): https://www.cia.gov/library/publications/the-world-factbook/index.html and https://www.cia.gov/library/publications/cia-maps-publications/index.html

NASA Earth Observatory (public domain): http://earthobservatory.nasa.gov/

Landsat: http://landsat.visibleearth.nasa.gov/

USGS EROS (Earth Resources Observatory and Science (EROS) Center) (public domain): http://eros.usgs.gov/#

Natural Earth (public domain): http://www.naturalearthdata.com/

[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: Partly

Reviewer #2: Yes

Reviewer #3: Partly

**********

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

Reviewer #1: I Don't Know

Reviewer #2: Yes

Reviewer #3: Yes

**********

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

Reviewer #3: 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

Reviewer #3: 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: The manuscript entitled “COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France” by Souris and Gonzalez has studied the influence of the hospital care system on lethality in metropolitan France during the inception of the COVID-19 epidemic. The study concludes that the higher case fatality rates observed in France is linked to the level of morbidity in the district. The study has also suggested that the case-fatality rates (CFR) may be linked to the hospital care available in the district, i.e., higher congestion of the hospitals or lower accessibility of the hospital care leads to the higher CFR. The manuscript is well written, the introduction is quite informative and the data are well presented and may be interesting for readers from varying fields. However, there are few concerns may be clarified are given below:

1. The abstract does not fully explains the study and needs to mention more about the findings presented in the discussion and conclusion.

2. Higher hospitalization leads to higher CFR, this explains that it is directly related to the higher morbidity, however, to compare with the hospital care system authors need to present a data showing the number of hospital/ beds accessible (per 1000). As explained higher hospital stress means 3 or more per 1000 does not explain the availability or accessibility to hospital care. Table 3 shows the total hospitalized and CFR. Here authors may put a data comparing the hospital care accessibility.

3. Table 3 also should show the data of hospitalized and death (%) 11 and 44 regions. This will help the readers to have better clarity.

4. The higher CFR is among the age group 50 and above. Is the overall difference found between CFR whole and CFR without regions 11 and 44 is due to the presence of higher number of higher age group cases?

5. Table 4 shows that the CFR varies from 0.134 for districts with a low hospitalization rate (less than 0.5 per 1000) to 0.150 for districts with a high hospitalization rate (less than 4 per 1000). How the average for all the districts higher (0.152)? What is the meaning of districts number? What does “all” in the last line shows?

6. Figure 12 presents districts with different morbidity and the lethality. Can the authors compare the low morbidity high lethality vs. high morbidity high lethality districts? Is there any similarity/ differences related to hospitalization/ accessible to hospitals or the aging population in these districts? The CFR is high in the higher age groups. To justify that the CFR is higher in the districts having higher hospital stress or due to lack of hospital care, authors need to show and compare the CFR rates between different age groups. Is the CFR is higher due to the presence of higher %age of higher age groups?

Reviewer #2: Comments to the Author

The authors show the increase in the case-fatality rate in France during epidemic peaks related to the level of morbidity, but not from the care or treatment of patients during hospital stress due to epidemic peaks. Although this paper was well written, I have several comments. Sufficient improvements for those issues are needed for the acceptance to the PLoS One.

Major comments

1. I believe that morbidity and mortality of COVID-19 are strongly related to patient’s factors including incidence of diabetes, and cardiovascular disease. Please indicate the difference of these risk factor between France and other countries or regions in France, or if there is no data, please indicate as a limitation.

2．In our country, there are differences in the availability of infectious disease wards and the capacity to operate ECMO depending on the region. Please explain differences of these factors between France and other countries or regions in France, or if there is no data please explain in the discussions and limitations.

3. Criteria of severity of COVID-19 by admitting ICU or not depend on capacity of each facility, so I concern that it may be inappropriate as an indicator of the severity. I consider that there is no correlation between the severity rate and the case-fatality rate. I believe that endotracheal intubation is relatively good indicator of the severity of the disease. Is it possible to discuss the difference in the severity of cases according to this standard and the difference from other countries?

Reviewer #3: Souris et al present a very unique and interesting analysis of the spatial distribution of COVID-19 cases requiring hospitalization in France, specifically analyzing hospital case-fatality rate to try to understand regional/geographic variations in healthcare. They report significant variations in CFR across regions, and conclude that higher CFR observed by districts are largely related to the level of morbidity/infection rate, and therefore to a putative overwhelming of the local healthcare systems during the acute phases of the epidemic, rather than any deficiency or failure of the local healthcare. Moreover, they suggest the geographic differences are not due to population risk factors or reporting systems, which they purport are uniform across France. Interestingly, the CFR variation is small compared to much greater variation across countries reported in the literature, suggesting marked international inconsistency in case reporting and death ascertainment/coding. The manuscript is very thorough, well-illustrated through figures, with very strong statistical assumptions-based robust analysis, largely well-written, and an important contribution at a time such as this.

Major Comments:

1. Abstract. Final sentence / concluding idea “However, the magnitude of this increase of case-fatality rate represents less than 10 per cent of the average case-fatality rate” is incomplete and leaves the findings unexplained; I would expand and clarify.

2. Data analysis.

a. (Pg 17): “The observed gross hospital case-fatality rate calculated by district showed a mean of 0.153 with a standard deviation of 0.045 … Without the Ile-de-France and Grand-Est regions, the mean gross case-fatality rate per district is 0.149 with a standard deviation of 0.048 (figure 6).” Given minimal change in mean CFR and SD by excluding the hardest hit areas, I don’t understand why the authors persist in the analysis with / without these regions

b. (Pg 18): “The spatial distribution of standardized morbidity rate (hospitalized cases) shows a significant spatial autocorrelation (Moran index: 1.54, p-value < 10-6), and this is expected for an infectious disease.” I don’t understand “autocorrelation” – a parameter with itself, or a value in a region correlating with values around it?

c. (Pg 19): “There is a correlation between the standardized hospitalization rate and the standardized case-fatality rate (Bravais-Pearson index=0.40) (Figure 10), a correlation which increases (0.48) if we limit the calculation to districts whose SLR is significantly different from 1 (p-value < 0.05).” It is unclear which correlation result the p value refers to? Relatedly, p value should be provided in Fig 10 or its legend.

3. Figures/Tables. Some are poorly formatted, with small fonts, poorly explained terminology, and with some sloppiness (French labels from software that are not translated/explained)

a. Fig 1. Y-axis title “Effectifs” unexplained. Why are data in decimal format compared to % format in Table 1?

b. Fig 2. Y-axis titles illegible. X-axis title “Calendar” not clearfor a linear numerical scale.

c. Table 3. Each age group is assigned a % for the various parameters, but the totals are decimal (1.0) !

d. Fig 12. The legend terminology “1-5, H for 1-4” is completely unclear! The wording is clear but does not relate to “H”.

4. Discussion: Several concluding statements are unclear, and don’t seem to be supported by data specifically presented in the Results. Moreover, much new data/analysis/calculations seem to be presented in the Discussion, which is inappropriate.

a. Pg 22: “the average case-fatality rate for all districts being about 15% higher than the average rate calculated in the 20% of districts with the lowest hospitalization rates” makes little sense. Moreover, the following conclusion “It is therefore very likely that the increase in hospital tension over the period under consideration has increased the hospital case-fatality rate” does not follow logically from the preceding.

b. Pg 23: “The first profile accounts for 60% of the patients (59,878 over 99,970). The case-fatality average for this profile is significantly higher (0.187) as compared with the other districts case-fatality average (0.138). The profile displays a 33% excess mortality, corresponding to 2,716 deaths (11,003 observed vs. 8,287 expected with the CFR 0.138). This excess mortality, corresponding to 16 % of all deaths, is likely to be the result of increased lethality due to overloading…” presents much more data/analysis than in the Results. Similarly for the 2nd profile.

c. Pg 24: “WHO data from 219 countries (WHO 20) show a mean case-fatality rate of 0.041 (median 0.029, standard deviation 0.045) with values between 0.31 (Nicaragua) and 0.00 (for countries that did not report deaths, like Vietnam). A T-test shows that the French average is therefore very significantly higher than the world average (p-value < 10-6)”

d. Pg 25: “difference observed between the average case-fatality rate in France and the average of the rates reported for all countries by international organizations or information sites (WHO, Wordometer, etc.). These differences probably stem from the reporting of cases and deaths, which is uneven from one country to another, and not from the care or treatment of patients” appears to be an unsubstantiated conclusion. Because the variation if pandemic stress in some regions of France only explained 10% of the overall CFR variation, why does this exclude the possibility that differences in healthcare quality/access could explain differences across countries?

5. Conclusions. Given the changing population of France with north African and other immigration, could regional population differences not explain some differences in risk of severity of COVID and thus CFR variation? Moreover, regarding the variation in European countries, population and healthcare quality differences are much too easily dismissed: (Pg 10) “In Europe, the characteristics of populations (in terms of risk factor for COVID-19) and health systems are comparable …”

6. References. I am very surprised given the complex geographic and statistical analysis that the reference list is so abbreviated ! It might be helpful to the average reader to perhaps have a few more to support the methodologic approaches, analysis, basics of healthcare statistics, etc.

Minor comments

1. Terminology.

a. Pg 16: “a cartographic representation by cartogram” should be revised.

b. Pg 21: “intensity of reanimation” is unclear.

c. Pg 22: “Hospital tension”

**********

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

Reviewer #3: Yes: Prof Sanjay Mehta, MDCM FRCPC

[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.

PLoS One. 2020 Dec 15;15(12):e0243606. doi: 10.1371/journal.pone.0243606.r002

Author response to Decision Letter 0

14 Aug 2020

Response to the reviewers

PONE-D-20-17129 COVID-19

From: Dr. Marc Souris, Directeur de Recherche, UMR 190 « Unités des virus émergents »

Object: Rebuttal letter to PONE-D-20-17129 COVID-19 review untitled “Spatial Analysis of Hospital Case-Fatality Rate in France” by Marc Souris and Jean-Paul Gonzalez.

To PLOS ONE Editor

Dear Editor,

We thank you to give us the opportunity with our manuscript to fully meet PLOS ONE’s publication criteria. You will find enclosed the revised version of the manuscript that addresses all the points raised by you and the reviewers. As requested, our re-submission includes the following items applied to the Editor’s requirements.

- The present letter as a rebuttal letter responding in detail to each point raised by the academic editor and reviewer(s).

- A marked-up copy of our manuscript with all changes made to the original version (file 'Revised Manuscript D-20-17129 with Track Changes'.

- A copy of our manuscript using track changes as been uploaded as a *supporting information* file.

- An unmarked version of our revised paper without tracked changes (file 'Manuscript D-20-17129 '. (Clean copy of the edited manuscript uploaded as the new *manuscript* file)

- We carefully pay attention for our manuscript to meet PLOS ONE's style requirements, including those for file naming.

- The entire manuscript has been reviewed for language usage, spelling, and grammar by a mother tongue English scientific speaker.

- The abstract has been reviewed and formatted for on the online submission form (via Edit Submission) and identical as the abstract in the manuscript.

- Copyright. We are perfectly aware and confirm that our Figure(s) 3, 4, 5, 6, 7, 8, 9, 11, 12 and 13 do not contain any copyright and our entirely produced by the authors of the present manuscript and will be published by PLOS under the Creative Commons Attribution License (CC BY 4.0) and acknowledge that the manuscript, images, and Supporting Information files will be freely available online, and any third party is permitted to access, download, copy, distribute, and use these materials in any way, even commercially, with proper attribution.

Please find below our answer addressing point by point all the comments raised by the reviewers.

Reviewer #1:

1. The manuscript entitled “COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France” by Souris and Gonzalez has studied the influence of the hospital care system on lethality in metropolitan France during the inception of the COVID-19 epidemic. The study concludes that the higher case fatality rates observed in France is linked to the level of morbidity in the district. The study has also suggested that the case-fatality rates (CFR) may be linked to the hospital care available in the district, i.e., higher congestion of the hospitals or lower accessibility of the hospital care leads to the higher CFR. The manuscript is well written, the introduction is quite informative, and the data are well presented and may be interesting for readers from varying fields.

Answer: We appreciate such positive comments and thank you for your consideration.

However, there are few concerns may be clarified are given below:

2. The abstract does not fully explain the study and needs to mention more about the findings presented in the discussion and conclusion.

Answer: The abstract has been completed as requested.

3. Higher hospitalization leads to higher CFR, this explains that it is directly related to the higher morbidity, however, to compare with the hospital care system authors need to present a data showing the number of hospital/ beds accessible (per 1000). As explained higher hospital stress means 3 or more per 1000 does not explain the availability or accessibility to hospital care. Table 4 shows the total hospitalized and CFR. Here authors may put a data comparing the hospital care accessibility.

Answer:

High hospitalization and therefore high morbidity should not lead to an increase in CFR, however one can think that excessive hospitalization (exceeding reception capacities) will potentially degrade hospital care.

The healthcare offer is designed to be evenly distributed in France: A table giving the number of beds per 100,000 inhabitants per region has been added, and the source of these data by departments. As it is indicated in the “results” section, there is no direct correlation between lethality and bed density, by department.

Accessibility does not only depend on the provision of care at regional or departmental level, but also depends on many other factors (e.g. socio-economic, behavioral, transport, rurality, etc.).

4. Table 4 also should show the data of hospitalized and death (%) 11 and 44 regions. This will help the readers to have better clarity.

Answer: The CFR for region 11 and 44 only has been added. However, the purpose of the study was to estimate the CFR by age group as a regular basis for the whole country, but not to analyze region 11 and 44 separately.

5. The higher CFR is among the age group 50 and above. Is the overall difference found between CFR whole and CFR without regions 11 and 44 is due to the presence of higher number of higher age group cases?

Answer: No, this is not true when the rate is age-standardized. In fact, the CFR estimate by age group for the whole country is used to standardize the rate on age by district, taking in account the population age structure of each district.

6. Table 4 shows that the CFR varies from 0.134 for districts with a low hospitalization rate (less than 0.5 per 1000) to 0.150 for districts with a high hospitalization rate (less than 4 per 1000). How the average for all the districts higher (0.152)? What is the meaning of districts number? What does “all” in the last line shows?

Answer: The table shows the increase of CFR with the increase of hospitalization rate. District number represent the cumulative number of districts corresponding to the hospitalization rate indicated in column 1.

7. Figure 12 presents districts with different morbidity and the lethality. Can the authors compare the low morbidity high lethality vs. high morbidity high lethality districts? Is there any similarity/ differences related to hospitalization/ accessible to hospitals or the aging population in these districts? The CFR is high in the higher age groups. To justify that the CFR is higher in the districts having higher hospital stress or due to lack of hospital care, authors need to show and compare the CFR rates between different age groups. Is the CFR is higher due to the presence of higher %age of higher age groups?

Answer: All rates are already age-standardized in order to take into account the difference of the age structure of a population between districts.

Reviewer #2:

8. The authors show the increase in the case-fatality rate in France during epidemic peaks related to the level of morbidity, but not from the care or treatment of patients during hospital stress due to epidemic peaks. Although this paper was well written, I have several comments. Sufficient improvements for those issues are needed for the acceptance to the PLoS One.

Answer: Yes, this true, only the case-fatality rate in France are presented during the epidemic peaks associated to the level of morbidity, by districts, but this is not an analysis done from the care or treatment of patients.

9. I believe that morbidity and mortality of COVID-19 are strongly related to patient’s factors including incidence of diabetes, and cardiovascular disease. Please indicate the difference of these risk factor between France and other countries or regions in France, or if there is no data, please indicate as a limitation.

Answer: In this study, the difference in CFR between districts in France is analyzed, and the magnitude of this difference is evaluated in order to compare it with the difference observed with CFR in other countries. The major observed risk factor for COVID-19 mortality is age, so the data has been age-standardized to exclude this factor as a potential difference between districts. Other risk factors for severity and mortality have been also observed for COVID-19, including among others: hypertension; cardiovascular disease history; diabetes; obesity. When trying to find another risk factor for SLR variation (as hospital congestion for example), it would also be useful to standardize on these health risk factors in order to eliminate differences in risk between districts. Unfortunately, apart for age, we do not have the necessary data to evaluate, for these risk factors or combinations of factors, the populations concerned (by districts) and the global case-fatality rates due to COVID-19 for each factor or a combination of factors. Even if these risk factors are also related to age, this is a limitation, and such appears now in the methods and discussion section.

10. In our country, there are differences in the availability of infectious disease wards and the capacity to operate ECMO depending on the region. Please explain differences of these factors between France and other countries or regions in France, or if there is no data please explain in the discussions and limitations.

Answer: As indicated in the section 8 (Q/A) above, our study analyzes the data reported by district without any clinical data on treatment or patient care at the hospital level. I our knowledge, no specific data on ECMO (Extracorporeal membrane oxygenation) capacity by district is available in France (only at the hospital level and not reported at the national level). By district, no data were available on treatment or care of patients (only for the intensive care units), therefore it was not possible to analyze the differences in lethality by referring to the patient treatment. This kind of analysis can be possible at the hospital level (providing such capacity) and eventually allow analyzing CFR differences between hospitals. But this will be another type of an interesting study.

11. Criteria of severity of COVID-19 by admitting ICU or not depend on capacity of each facility, so I concern that it may be inappropriate as an indicator of the severity. I consider that there is no correlation between the severity rate and the case-fatality rate. I believe that endotracheal intubation is relatively good indicator of the severity of the disease. Is it possible to discuss the difference in the severity of cases according to this standard and the difference from other countries?

Answer: We agree with the comment of the reviewer. In fact, we think that in France ICU admission for COVID patients means most of the time, respiratory distress followed by endotracheal intubation. However, as a criteria of severity of COVID-19 by admitting ICU or not depend not only on the capacity of each facility (some have high capacity and generally before and after the epidemic pic) but also on the criteria (score) used by the ICU team to evaluate the patient (e.g. NEW2) and also depends on the country. Unfortunately, we don’t have access to such type of data for other countries. In this article, we use “severity” based only the rate of resuscitation measures or endotracheal intubation.

Reviewer #3

12. Souris et al present a very unique and interesting analysis of the spatial distribution of COVID-19 cases requiring hospitalization in France, specifically analyzing hospital case-fatality rate to try to understand regional/geographic variations in healthcare. They report significant variations in CFR across regions, and conclude that higher CFR observed by districts are largely related to the level of morbidity/infection rate, and therefore to a putative overwhelming of the local healthcare systems during the acute phases of the epidemic, rather than any deficiency or failure of the local healthcare. Moreover, they suggest the geographic differences are not due to population risk factors or reporting systems, which they purport are uniform across France. Interestingly, the CFR variation is small compared to much greater variation across countries reported in the literature, suggesting marked international inconsistency in case reporting and death ascertainment/coding. The manuscript is very thorough, well-illustrated through figures, with very strong statistical assumptions-based robust analysis, largely well-written, and an important contribution at a time such as this.

Answer: We appreciate the point of view of the reviewer and entirely agree with this overview of our work. Thank you.

13. Abstract. Final sentence / concluding idea “However, the magnitude of this increase of case-fatality rate represents less than 10 per cent of the average case-fatality rate” is incomplete and leaves the findings unexplained; I would expand and clarify

Answer: As above mentioned, the abstract has been entirely revised accordingly to the positive suggestion and comments by the reviewers.

14. Data analysis.

Answer: Indeed, the difference is not very significant, but it seemed to us more rigorous to standardize on age with overall rates estimated without a possible increase due to a possible saturation of the care system.

Answer: Yes, spatial autocorrelation measures the correlation of a value with the values around it and reflect the spatial pattern of a phenomenon. References has been added.

Answer: The p-value refers to the significance of the SLR to be different from 1, i.e. the district showing statistically significant excess or default of SLR. The correlation can be calculated over all districts, or only over the districts that show statistically significant excess or default lethality.

15. Figures/Tables. Some are poorly formatted, with small fonts, poorly explained terminology, and with some sloppiness (French labels from software that are not translated/explained)

Answer: All French labels has been translated accordingly. All figures and tables have been revised.

a. Fig 1. Y-axis title “Effectifs” unexplained. Why are data in decimal format compared to % format in Table 1?

Answer: Corrected

b. Fig 2. Y-axis titles illegible. X-axis title “Calendar” not clear for a linear numerical scale.

Answer: Corrected

c. Table 3. Each age group is assigned a % for the various parameters, but the totals are decimal (1.0) !

Answer: Corrected

d. Fig 12. The legend terminology “1-5, H for 1-4” is completely unclear! The wording is clear but does not relate to “H”.

Answer: Corrected

16. Discussion: Several concluding statements are unclear, and don’t seem to be supported by data specifically presented in the Results. Moreover, much new data/analysis/calculations seem to be presented in the Discussion, which is inappropriate.

Answer: We reformulate and clarified the sentence.

Answer: We have chosen to place these results in the discussion because they are part of the analysis of the different profiles we are proposing, which are more a matter of discussion than of results per se.

Answer: We moved this paragraph to the “results” section.

Answer: The healthcare system in France is renowned for being of good quality, in terms of equipment, medical staff, patient care and treatment. COVID-19 patient management protocols in France do not differ significantly from those used in other European countries (Germany, Spain, Italy, Greece, etc.). It is therefore unlikely that the huge differences in lethality are due to differences in management or quality of care. But our conclusion concerns the possible degradation of care during periods of hospital stress (epidemic pic, ward and medical worker overload): the results indicate that this observed degradation, even substantial, cannot explain the difference in case-fatality rate with other countries, some of which have themselves experienced periods of hospital stress (e.g. Italy, Spain, UK, Mexico, Brazil).

17. Conclusions. Given the changing population of France with north African and other immigration, could regional population differences not explain some differences in risk of severity of COVID and thus CFR variation? Moreover, regarding the variation in European countries, population and healthcare quality differences are much too easily dismissed: (Pg 10) “In Europe, the characteristics of populations (in terms of risk factor for COVID-19) and health systems are comparable …”.

Answer: Official data collection and statistics on ethnic origin of population are prevented in France.

18. References. I am very surprised given the complex geographic and statistical analysis that the reference list is so abbreviated! It might be helpful to the average reader to perhaps have a few more to support the methodologic approaches, analysis, basics of healthcare statistics, etc.

Answer: Several references have been added accordingly.

19. Terminology.

a. Pg 16: “a cartographic representation by cartogram” should be revised.

Answer: “Cartogram” is the specific name of this king of map, in which some thematic variable is substituted for land area or distance.

b. Pg 21: “intensity of reanimation” is unclear.

Answer: we change to “the differences in use of resuscitative measures or endotracheal intubation”

c. Pg 22: “Hospital tension”

Answer: we change to “high medical admission rates and healthcare workers stress”

Dear Editor, we thank you for your kind consideration of our reviewed manuscript.

Dr. Marc Souris

Dr. Jean-Paul Gonzalez

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(150.5KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0243606.r003

Decision Letter 1

Muhammad Adrish

1 Sep 2020

PONE-D-20-17129R1

COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France

PLOS ONE

Dear Dr. Souris,

==============================

ACADEMIC EDITOR: Please see comments by the reviewers. I am requesting you to address these concerns before I can accept the study for publication.

==============================

Please include the following items when submitting your revised manuscript:

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'.

We look forward to receiving your revised manuscript.

Kind regards,

Muhammad Adrish

Academic Editor

PLOS ONE

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

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

Reviewer #3: (No Response)

**********

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

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

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

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: I Don't Know

**********

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

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

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

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Reviewer #1: The authors have satisfactorily answer all the queries raised during revision. The manuscript now may be considered for publication.

Reviewer #2: The authors show the increase in the case-fatality rate in France during epidemic peaks related to the level of morbidity during hospital stress due to epidemic peaks. Author had responded correctly to my queries.

Reviewer #3: I thank the authors for extensive revisions to the paper, including improving the figures, legends, tables and abstract. The paper reads better. There are however a few issues I raised which were too easily dismissed and perhaps my concern not clear.

Major Comments:

1. My previous comment: d. Pg 25: “difference observed between the average case-fatality rate in France and the average of the rates reported for all countries by international organizations or information sites (WHO, Wordometer, etc.). These differences probably stem from the reporting of cases and deaths, which is uneven from one country to another, and not from the care or treatment of patients” appears to be an unsubstantiated conclusion. Because the variation if pandemic stress in some regions of France only explained 10% of the overall CFR variation, why does this exclude the possibility thatdifferences in healthcare quality/access could explain differences across countries?

Unresolved concern: The crux of the paper is that “epidemic stress” leading to some degradation of care explains only a small proportion of the variation in CFR across French regions. However, the authors have not clearly discussed or speculated what is the major reason for this huge variation in lethality! Moreover, their statements of “good quality” of French care and similar patient management protocols across European countries ignores the reality of medical care at the frontline or on the ground, where much variation occurs based on local administrations, local facilities, as well as local populations. Their response is inadequate.

2. My previous comment: Conclusions. Given the changing population of France with north African and other immigration, could regional population differences not explain some differences in risk of severity of COVID and thus CFR variation? Moreover, regarding the variation in European countries, population and healthcare quality differences are much too easily dismissed: (Pg 10) “In Europe, the characteristics of populations (in terms of risk factor for COVID-19) and health systems are comparable …”.

Answer: Official data collection and statistics on ethnic origin of population are prevented in France.

Unresolved concern: This limitation should be recognized/stated in the manuscript, but is not an excuse for lack of scientific consideration by the authors of this important basis for some of the regional / country differences. This point should be considered/discussed even in the absence of any specific data being available.

Minor comments

1. My previous comment: Pg 16: “a cartographic representation by cartogram” should be revised.

Answer: “Cartogram” is the specific name of this king of map, in which some thematic variable is substituted for land area or distance.

Unresolved concern: I accept the term is correct, but the language is awkward/repetitious, “… cartographic representation by cartogram” !

**********

7. 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: Yes: Masaki Okamoto

Reviewer #3: Yes: Professor Sanjay Mehta, MDCM FRCPC

PLoS One. 2020 Dec 15;15(12):e0243606. doi: 10.1371/journal.pone.0243606.r004

Author response to Decision Letter 1

8 Sep 2020

Please find below our answer addressing the comments raised by the reviewer.

Reviewer #3

R#3- Unresolved concern: The crux of the paper is that “epidemic stress” leading to some degradation of care explains only a small proportion of the variation in CFR across French regions. However, the authors have not clearly discussed or speculated what is the major reason for this huge variation in lethality! Moreover, their statements of “good quality” of French care and similar patient management protocols across European countries ignores the reality of medical care at the frontline or on the ground, where much variation occurs based on local administrations, local facilities, as well as local populations. Their response is inadequate.

Answer:

When we refer strictly to the data, the standard deviation of the case-fatality rate between French departments is 4.5%, for a case-fatality rate that averages 15.3%. Our results show that these variations are directly related to the level of morbidity in the department, and cannot be explained by possible differences between departments in the COVID-19 assumed risk factors for lethality (age, cardiovascular history, hypertension, diabetes, obesity).

Independently of the causes of the internal spatial variability of lethality in France, we therefore find that this internal spatial variability is too low to explain the significant differences with the rates published for many other countries: the differences in mean lethality between countries cannot be explained either by the differences in morbidity between countries, or by the differences in lethality risk factors between countries, or by the differences in inpatient treatment (at the global mean level. We certainly do not ignore the existing disparities in the reality of the field, but in this study, it is not a question of going into detail at the granular level of the hospital).

As we write in the article, the most likely explanation for these strong differences in lethality between countries is therefore the disparities in the reporting of cases and deaths. This explanation is confirmed with the publication of studies on excess mortality, which show that most countries have a deficit of deaths (not enough COVID-19 deaths reported to explain excess mortality compared to previous years), while some countries, including France and Belgium, have a slight excess of COVID-19 deaths (too many COVID-19 deaths reported compared to excess mortality compared to previous years).

In order to clarify our point, we have added these few remarks to the end of the discussion (p. 21) and added references to the studies mentioned above.

R#3-Unresolved concern: This limitation should be recognized/stated in the manuscript, but is not an excuse for lack of scientific consideration by the authors of this important basis for some of the regional / country differences. This point should be considered/discussed even in the absence of any specific data being available.

Answer :

Indeed, our answer was probably incomplete.

Beyond the technical difficulty of invoking ethnic differences or standardizing on this variable (due to lack of data), we consider that ethnic, religious, cultural factors are confounding factors (in the epidemiological sense of the term), in that they can influence the COVID-19 risk factors of lethality (e. g. co-morbidities, access to care, late recourse to care) but are not directly risk factors of lethality themselves: data on lethality risk factors related to the health status of individuals are available, and we already had added a paragraph on this issue in the last revision of the manuscript (page 15 in the Results section).

Recent studies on excess mortality in March-April 2020 show differences (globally over France) according to the geographical origin of birth of the deceased (but beware: it is not known whether the death is attributable to COVID-19). These differences show an excess of excess mortality among individuals born abroad, particularly in Africa and Asia. The authors of these studies attribute this excess of excess mortality due to COVID-19 to the spatial distribution of this population, which is much more present in densely populated areas, particularly in the Ile-de-France region, and which has therefore been proportionally more affected by the epidemic. These studies concern morbidity and mortality, not lethality, and do not suggest any risk factors for lethality specific to this population. No genetic factor specific to ethno-geographical origin has been evoked to be associated with COVID-19 lethality rates. As suggested by the reviewer, we add a paragraph in the discussion to refer to this question (page 20).

R#3 Minor comments -Unresolved concern: I accept the term is correct, but the language is awkward/repetitious, “… cartographic representation by cartogram” !

Answer: It has been adjusted accordingly in the Figure title by deleting the « cartographic » and rewording the sentence.

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(150.5KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0243606.r005

Decision Letter 2

Muhammad Adrish

23 Oct 2020

PONE-D-20-17129R2

COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France

PLOS ONE

Dear Dr. Souris,

==============================

ACADEMIC EDITOR: Please review the comments made by the reviewers and provide a point by point response in your revised manuscript.

==============================

Please include the following items when submitting your revised manuscript:

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'.

We look forward to receiving your revised manuscript.

Kind regards,

Muhammad Adrish

Academic Editor

PLOS ONE

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

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #3: All comments have been addressed

Reviewer #4: (No Response)

**********

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

Reviewer #3: Yes

Reviewer #4: No

**********

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

Reviewer #3: I Don't Know

Reviewer #4: No

**********

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

Reviewer #3: Yes

Reviewer #4: No

**********

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

Reviewer #3: Yes

Reviewer #4: No

**********

6. Review Comments to the Author

Reviewer #3: (No Response)

Reviewer #4: 1) The concrete results in the manuscript include spatial autocorrelation of certain variables, a positive correlation between hospitalization rate and case-fatality rate, and a “weak” negative correlation between hospitalization rate and severity rate. However, there is concrete data and rigorous analysis to support one point the authors tried to stress—non-comparability of case-fatality rates across countries. Therefore, statements on this point should at most be kept in Discussion and removed from Abstract (the 4th paragraph) and Introduction (the 4th paragraph).

2) The exact definition of age standardized case-fatality rate is still unclear. In the formula in page 4, are P[a] and T[a] district specific or not? If they are, there should be subscript i; if not, that means the expectation E is the same in all the district of the France, which is the so-called “average rate for France”? It is unclear whether the district observed case fatality rate was age adjusted or not? My understanding is yes according to the term “district’s SLR”; however, what’s the formula for it? As I pointed out, the only formula so far is not district specific.

3) The language is verbose and many contents are superfluous. I suggest shortening the manuscript, removing unnecessary parts, and making only succinct points. Take an example, the whole page 4 seems can be removed.

4) The last sentence of the second paragraph in Introduction—“…, unlike case-fatality rates, which are normally calculated independently of the number of infected persons”— it is confusing to me, how the case-fatality rate could be independent of the number of infected persons? Isn’t it the denominator?

**********

7. 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 #3: Yes: Professor Sanjay Mehta, MDCM FRCPC

Reviewer #4: No

PLoS One. 2020 Dec 15;15(12):e0243606. doi: 10.1371/journal.pone.0243606.r006

Author response to Decision Letter 2

23 Oct 2020

Please find below our answers addressing the comments raised by the reviewer#4.

1) The concrete results in the manuscript include spatial autocorrelation of certain variables, a positive correlation between hospitalization rate and case-fatality rate, and a “weak” negative correlation between hospitalization rate and severity rate. However, there is concrete data and rigorous analysis to support one point the authors tried to stress—non-comparability of case-fatality rates across countries. Therefore, statements on this point should at most be kept in Discussion and removed from Abstract (the 4th paragraph) and Introduction (the 4th paragraph).

Answer: we removed the statement of the introduction (effectively it is a result of our study), but we think that it can be kept as a result in the abstract. We move the statement at the end of the abstract to clearly show it as a conclusion.

Answer: Indirect age standardization use case-fatality rates by age group a (Ta) calculated over the whole country, and not district specific, but Pa,i is the population of age group a in the district i. The subscript i was missing for Pa,i. We corrected this error, and we thank the reviewer to have highlighted this error in the formula.

Answer: we agree that the manuscript is long, but we think that explanations on lethality and its different factors are required to better understand the article (as noted positively by the other reviewers).

Answer: The sentence meaning is: the case-fatality rate is independent of the extend of the disease in the population. We removed the sentence to avoid confusion.

5) I confirm that all the data used in this study is fully available, as indicated in the article.

Attachment

Submitted filename: Response to Reviewers_3.docx

Click here for additional data file.^{(16.9KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0243606.r007

Decision Letter 3

Muhammad Adrish

3 Nov 2020

PONE-D-20-17129R3

COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France

PLOS ONE

Dear Dr. Souris,

==============================

ACADEMIC EDITOR: Please see below the unaddressed concern raised by the reviewer. I am unable to accept your study until these concerns have been appropriately addressed.

==============================

Please include the following items when submitting your revised manuscript:

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'.

We look forward to receiving your revised manuscript.

Kind regards,

Muhammad Adrish

Academic Editor

PLOS ONE

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

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #4: All comments have been addressed

**********

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

Reviewer #4: Yes

**********

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

Reviewer #4: Yes

**********

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

Reviewer #4: Yes

**********

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

Reviewer #4: Yes

**********

6. Review Comments to the Author

Reviewer #4: 1) The authors currently put in the last sentence of Abstract “…, suggesting marked international inconsistency in case reporting and death ascertainment/coding”, which is exactly what I commented—it is an implication from the results, not the direct results. Note that it is a very strong statement. Therefore, please remove it from Abstract and leave it only in Discussion.

**********

7. 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 #4: No

PLoS One. 2020 Dec 15;15(12):e0243606. doi: 10.1371/journal.pone.0243606.r008

Author response to Decision Letter 3

3 Nov 2020

Answer: we removed the statement from the abstract.

Attachment

Submitted filename: Response to Reviewers_4.docx

Click here for additional data file.^{(15.1KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0243606.r009

Decision Letter 4

Muhammad Adrish

25 Nov 2020

COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France

PONE-D-20-17129R4

Dear Dr. Souris,

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,

Muhammad Adrish

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

You have satisfactorily answered all queries made by the reviewers

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #4: All comments have been addressed

**********

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

Reviewer #4: Yes

**********

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

Reviewer #4: Yes

**********

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

Reviewer #4: Yes

**********

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

Reviewer #4: Yes

**********

6. Review Comments to the Author

Reviewer #4: My comments were all addressed, I have no further critique.

My comments were all addressed, I have no further critique.

**********

7. 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 #4: No

PLoS One. doi: 10.1371/journal.pone.0243606.r010

Acceptance letter

Muhammad Adrish

2 Dec 2020

PONE-D-20-17129R4

COVID-19: Spatial Analysis of Hospital Case-Fatality Rate in France

Dear Dr. Souris:

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. Muhammad Adrish

Academic Editor

PLOS ONE

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(150.5KB, docx)}

Attachment

Submitted filename: Response to Reviewers.docx

Click here for additional data file.^{(150.5KB, docx)}

Attachment

Submitted filename: Response to Reviewers_3.docx

Click here for additional data file.^{(16.9KB, docx)}

Attachment

Submitted filename: Response to Reviewers_4.docx

Click here for additional data file.^{(15.1KB, docx)}

Data Availability Statement

[pone.0243606.ref001] 1.Morteza AK, Fakher R. Cross-Country Comparison of Case Fatality Rates of COVID-19/SARS-COV-2. Osong Public Health and Research Perspectives 2020; 11(2): 74–80. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0243606.ref002] 2.Rinaldia G, Paradisiba M. An empirical estimate of the infection fatality rate of COVID-19 from the first Italian outbreak. Harvard Economics Department. [Google Scholar]

[pone.0243606.ref003] 3.Spychalski P, Blazynska-Spychalska A, Kobiela J. Estimating case fatality rates of COVID-19. The Lancet, 10.1016/S1473-3099(20)30246-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0243606.ref004] 4.Baud D, Qi X, Nielsen-Saines K, Musso D, Pomar L, Favre G. Real estimates of mortality following COVID-19 infection. Lancet Infect Dis. 2020. (published online March 12). 10.1016/S1473-3099(20)30195-X [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0243606.ref005] 5.Porta M. A dictionary of Epidemiology. 5th Ed Oxford University Press, Oxford, 2008. [Google Scholar]

[pone.0243606.ref006] 6.Collier L, Oxford J. Human Virology. Oxford University Press, 2016. [Google Scholar]

[pone.0243606.ref007] 7.Saluzzo JF, Vidal P, Gonzalez JP. Les virus émergents. IRD Edition, 2004. [Google Scholar]

[pone.0243606.ref008] 8.Hurraux JM, Nicolas JC, Agut H, Peigne-Lafeuille H. Traité de virologie humaine. Estem/AUF, 2003. [Google Scholar]

[pone.0243606.ref009] 9.Salie H, Tran-Kiem C, Lefrancq N, Courtejoie N, Bosetti P, et all. Estimating the burden of SARS-CoV-2 in France. 2020. Science 13 May 2020. 10.1126/science.abc3517 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0243606.ref010] 10.Bouyer J, Hémon D, Cordier S, Derriennic F, Stûcker I, Stengel B, et al. Epidémiologie, principes et méthodes quantitatives. Inserm, 1993. [Google Scholar]

[pone.0243606.ref011] 11.Breslow NE and Day NE, Statistical methods in cancer research, vol. I & II. WHO, International Agency for Reseach on Cancer, 1987. [Google Scholar]

[pone.0243606.ref012] 12.Samuels SJ, Beaumont JJ, Breslow NE (1991). Power and detectable risk of seven tests for SMR. Am J Epidemiology 1991; 133:1191–1197. [DOI] [PubMed] [Google Scholar]

[pone.0243606.ref013] 13.Miller RG. Simultaneous Statistical Inference. Springer Verlag, 1996. [Google Scholar]

[pone.0243606.ref014] 14.Souris M. Epidemiology and Geography. Principles, methods and tools of spatial analysis. Wiley-ISTE, London, 2019 [Google Scholar]

[pone.0243606.ref015] 15.Pfeiffer D, Robinson T, Stevenson M, Stevens K, Rogers D, Clements A. Spatial Analysis in Epidemiology. Oxford University Press, 2008. [Google Scholar]

[pone.0243606.ref016] 16.Allard T, Bayardin V, Mosny E. L’Île-de-France, région la plus touchée par le surcroît de mortalité pendant le confinement. Insee Analyses Île-de-France n° 118, Juin 2020. [Google Scholar]

[pone.0243606.ref017] 17.Papon S, Robert-Bobée I. Une hausse des décès deux fois plus forte pour les personnes nées à l’étranger que pour celles nées en France en mars-avril 2020. Insee Focus No 198—Juillet 2020 [Google Scholar]

[pone.0243606.ref018] 18.Le Temps. L’impact du Covid-19 sous-estimé dans de nombreux pays. Available online: https://labs.letemps.ch/interactive/2020/surmortalite-et-covid-19/

PERMALINK

COVID-19: Spatial analysis of hospital case-fatality rate in France

Marc Souris

Jean-Paul Gonzalez

Roles

Abstract

Introduction

Table 1. Case-fatality rates and characteristics of countrywide population vulnerability to COVID-19.

Fig 1. Distribution of case-fatality rates calculated by country, based on case and death data from WHO (WHO, May 8, 2020).

Table 2. SARS-Cov-2 detection, COVID-19 morbidity and mortality rates by selected country.

Data and methods

Data

Fig 2.

Fig 3. The 96 districts of metropolitan France.

Table 3. Hospital bed number per 100,000 people, per regions in France.

Methods

Age standardization

Standardized case-fatality rate

Cartography

Study of statistical correlations

Results

Table 4. Hospitalization, mortality and case-fatality rates by age group in metropolitan France (2020).

Fig 4.

Fig 5. Cartogram of the standardized hospitalization rate for COVID-19 in France from March 19 to May 8, 2020.

Fig 6.

Fig 7. Calculated case-fatality rates by 13-day period.

Fig 8. Standardized Lethality Ratio (SLR) for COVID-19 in metropolitan France for the period March 19 to May 8, 2020.

Fig 9.

Fig 10. Correlation between hospitalization rate and CFR during the COVID-19 epidemic in France for the period of time between March 19 to May 8, 2020.

Table 5. Average standardized CFR as a function of hospitalization rate during the COVID-19 epidemic in France for the period of time between March 19 to May 8, 2020.

Fig 11.

Fig 12. Combination of standardized hospitalization (H) and case-fatality (L) rates in five classes.

Fig 13. Severity ratio during the COVID-19 epidemic in France for the period of time between March 19 to May 8, 2020.

Discussion

Conclusion

Data Availability

Funding Statement

References

Decision Letter 0

Muhammad Adrish

Roles

Author response to Decision Letter 0

Decision Letter 1

Muhammad Adrish

Roles

Author response to Decision Letter 1

Decision Letter 2

Muhammad Adrish

Roles

Author response to Decision Letter 2

Decision Letter 3

Muhammad Adrish

Roles

Author response to Decision Letter 3

Decision Letter 4

Muhammad Adrish

Roles

Acceptance letter

Muhammad Adrish

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases