Correlation of the rise and fall in COVID-19 cases with the social isolation index and early outpatient treatment with hydroxychloroquine and chloroquine in the state of Santa Catarina, southern Brazil: A retrospective analysis.

BACKGROUND: To contribute to the understanding of the coronavirus disease (COVID-19) pandemic, this study evaluated the correlations of the frequencies of COVID-19 cases, hospitalisations due to COVID-19, and deaths due to COVID-19 with social isolation indices and outpatient prescriptions of hydroxychloroquine and chloroquine in the state of Santa Catarina, southern Brazil.
METHODS: This was an analytical, observational, retrospective study based on secondary data that were obtained from public Brazilian databases and covered the period from March 1, 2020 to October 31, 2020 (epidemiological weeks 10-44). Data on weekly COVID-19 cases, hospitalisations and deaths due COVID-19, sales of chloroquine and hydroxychloroquine, and social isolation indices were obtained. Associations between the variables were tested using multiple linear regression analysis.
RESULTS: In all regions of Santa Catarina, there were almost simultaneous peaks of COVID-19 pandemic in weeks 28-31, followed by a sudden decrease. Social isolation indices were not associated with the outcomes; sales of chloroquine and hydroxychloroquine were significant predictors of all outcomes (p < 0.001). COVID-19 prevalence was significantly different across the state regions when COVID-19 cases started to decline (p < 0.001). DISCUSSION: Collective immunity and social isolation may not have been the only causes for the reduction of the COVID-19 pandemic observed in Santa Catarina. The results of this study were compatible with the hypothesis that early treatment of COVID-19 cases with chloroquine or hydroxychloroquine may contribute to reducing the transmissibility of COVID-19 in the population. This hypothesis needs to be further tested in future studies.

Introduction

Coronavirus disease (COVID-19) is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and emerged in December 2019 in Wuhan, China [1]. The outbreak was declared a public health emergency of international concern by the World Health Organization (WHO) on January 30, 2020. The Federal Government of Brazil, the largest country in South America, declared COVID-19 a national public health emergency on February 3, 2020 [2]. The first confirmed COVID-19 case in Brazil was reported on February 25, 2020 in a traveller returning to São Paulo from northern Italy [3]. Following this, the disease spread rapidly throughout the country. On March 11, 2020, the WHO declared COVID-19 a pandemic [4].

Although the mortality of the disease caused by SARS-CoV-2 is lower than that of diseases caused by other coronaviruses, its high transmissibility rapidly led to more frequent hospitalisations due to severe acute respiratory syndrome (SARS) and more deaths in terms of absolute numbers than previous respiratory epidemics [5]. In many countries, over a few weeks, excessive hospital overloading and extreme shortages of healthcare resources led to the “collapse” of local healthcare systems [6].

Since human-to-human transmission occurs primarily via respiratory droplets and direct contact, non-pharmacological interventions at the population level, such as social distancing, were indicated for the mitigation of COVID-19 spread at local and global levels [7]. In Brazil, a federative republic composed of 26 states and a federal district, all federative units implemented social distancing laws, mostly early during the pandemic (between the first and the tenth case of COVID-19 in its area), in the latter half of March 2020 [8].

However, the prolonged duration of the pandemic and economic impact of social isolation have led to a progressive reduction in adherence to these rules in the population, as well as the reduction of restrictions mandated by government authorities. In parallel, initiatives emerged that offered the possibility of off-label pharmacological treatment for COVID-19 as a pandemic control measure, despite negative or inconclusive results in a few clinical trials regarding outpatient treatment of COVID-19 cases [9]. Based mainly on positive results from observational studies [10] and previous positive in vitro findings [11], several early treatment protocols were propagated in Brazil, supported by the Federal Ministry of Health, with off-label use of hydroxychloroquine (HCQ) or chloroquine (CQ) as the main therapeutic agent [12]. However, most public health authorities in Brazilian states did not support the use and distribution of these medications through the Brazilian public health system, Sistema Único de Saúde (SUS). As a result, federal guidance for the early use of HCQ or CQ was not implemented by the public health system in most Brazilian cities. The use of COVID-19 early treatment protocols and the prescription of these drugs were based on the individual decisions of healthcare practitioners. Access to medications was achieved mainly through private drugstores. In Brazil, private health establishments are allowed to operate independently and are considered complementary to the public health system.

This led to a complex, mixed scenario of partial adherence to social distancing measures and limited access to pharmacological therapy. The relationship between the prescription of early pharmacological therapy, adherence to social isolation, and progression of the pandemic has not yet been studied at the populational level.

To contribute to the understanding of the evolution of the pandemic in Brazil, this study aimed to describe the correlations of the frequencies of COVID-19 cases, hospitalisations due to COVID-19, and deaths due to COVID-19 with the social isolation rates and number of outpatient prescriptions for HCQ and CQ in the state of Santa Catarina, southern Brazil. For comparative purposes, the same data were obtained for the neighbouring Brazilian state of Paraná. The evolution of the pandemic in Santa Catarina is well-documented, with specific data on COVID-19 cases and deaths due to COVID-19 in each of the state sanitary subdivisions (health macro-regions).

Material and methods

An analytical, observational, retrospective study was carried out based on secondary data obtained from public Brazilian databases that covered the period from March 1, 2020 to October 31, 2020 (epidemiological weeks 10–44). This study analysed data from the states of Santa Catarina and Paraná; these were chosen because they are two neighbouring states in the same region of Brazil (southern region), with similar geographical, demographic, and economic characteristics.

In Brazil, all COVID-19 cases have been reported on the e-SUS Epidemiological surveillance – Vigilância Epidemiológica (e-SUS-VE), a national electronic COVID-19 reporting system, and hospitalised COVID-19 cases with severe acute respiratory syndrome were recorded in the pre-existing Influenza Epidemiological Surveillance Information System (Sistema de Informação de Vigilância Epidemiológica da Gripe (SIVEP-Gripe) system. The SIVEP-Gripe system has been used since 2009 (initially implemented in response to the 2009 influenza H1N1 pandemic) and has since nationally centralised the reporting of hospitalisations due to any respiratory virus. Both the e-SUS-VE and SIVEP-Gripe include suspected and confirmed COVID-19 cases reported by public health and private services. These two reporting systems are interrelated in the Brazilian COVID-19 portal (https://covid.saude.gov.br/) [13]. For this study, data on hospitalisations and deaths due to COVID-19 were obtained from the analysis of the complete spreadsheet (csv file) of the National Severe Acute Respiratory Syndrome Database, made available by the COVID-19 portal; this was accessed on November 16, 2020 [14]. The cases retrieved from this file were kept only if the variable “final case classification” (CLASSI_FIN) was defined as “due to COVID-19" (value “5" '). From this file containing data from all over Brazil, only case records from the states of Santa Catarina (SC) and Paraná (PR) were included in this study, based on the variable “Acronym of the Federation Unit” (SG_UF_NOT). The cases in each state were organised from oldest to most recent, according to the variable “Epidemiological Week of the First Symptoms” (SEM_PRI) [15].

Brazilian states also have local epidemiological surveillance systems. Data on all COVID-19 cases were more detailed in these local databases than in the national database. Thus, data on the total number of COVID-19 cases registered in the states of Santa Catarina and Paraná were obtained from complete spreadsheets (csv files) made available by the respective state health departments [16,17]. The cases in each state were organised from oldest to most recent, based on the variable “date of onset of symptoms” (data_início_sintomas) [18], and grouped according to the epidemiological weeks. To avoid including errors in the date of symptom onset, cases were excluded from the study if the date of symptom onset was not recorded, or when the registered symptom onset date was later than the date of confirmation of diagnosis or more than thirty days before the date of confirmation of the diagnosis. To assess the impact of these exclusion criteria, COVID-19 cases were also selected for a complementary analysis based on the date of confirmation of diagnosis instead of the date of onset of symptoms. For this analysis, in the data sheet of Santa Catarina, the variable “date of collection” (data_coleta) was used, and in the data sheet of Paraná, the variable “date of diagnosis” (data_diagnóstico) was used [18].

To allow comparisons between the results of Santa Catarina and Paraná, all the values regarding hospitalisations due to COVID-19, deaths due to COVID-19, and COVID-19 cases were normalised as a proportion in relation to the state's population as estimated in July 2020 [19].

The social isolation indices in the states of Santa Catarina and Paraná were obtained from the “Brazilian COVID-19 Map” (2020 Inloco©, Recife, Brazil) [20]. According to the company, data on the locational behaviour of individuals are obtained through a network of mobile apps that, upon the user's consent, make the user's location information available to the company in an anonymised and aggregated manner. The company data bank comprises approximately 60 million Brazilian cell phone users (about 28.3% of the population) [19]. The social isolation index was then calculated as the number of users in a region that did not leave their place of residence on a given day in relation to the total number of users in the same region. For this study, the mean social isolation index recorded for seven days a week, from Sunday to Saturday, was considered as the social isolation index for that specific epidemiological week.

The number of units (capsules or pills) of HCQ and CQ sold for outpatient use in the states of Santa Catarina and Paraná were obtained by consulting the National System for the Management of Controlled Products of the National Health Surveillance Agency - ANVISA [21]. The system makes the number of units of a certain drug sold per month for each Brazilian state available. Data from March to October were collected for the states of Santa Catarina and Paraná. For each state, the total number of units sold (the sum of capsules and pills) of all HCQ and CQ formulations, industrialised or manipulated, was recorded for each month. Liquid preparations were not considered. The average number of units sold per epidemiological week was estimated by dividing the number of units sold in the month by either four or five, depending to the number of weeks in the month. Data on HCQ and CQ sales were also normalised in relation to the state population as estimated in July 2020 [19].

Demographic, geographic, and economic data of the states of Santa Catarina and Paraná were obtained from the Brazilian Institute of Geography and Statistics - IBGE - portal [22]. Analysis of the results in Santa Catarina was detailed in the state sanitary subdivisions, called “health macro-regions”. The definition of macro-regions inside the state of Santa Catarina and population of each macro-region were considered based on the most recent definitions by the Santa Catarina State Secretary of Health [23].

Data were stored and analysed using Excel 365 (Microsoft©, Redmond, USA). Statistical analyses were performed using IBM SPSS Statistics for Windows version 27.0 (IBM Corp.©, Armonk, USA). Comparisons between the mean results of weekly hospitalisations due to COVID-19, deaths due to COVID-19, COVID-19 cases, social isolation indices, and sales of HCQ and CQ in the states of Santa Catarina and Paraná were made using the paired Student's t-test. The associations between the outcome variables (hospitalisations due to COVID-19, deaths due to COVID-19, COVID-19 cases) and predictor variables (social isolation indices and sales of HDQ and CQ) were evaluated using a multiple linear regression model with hierarchical entry of predictor variables (predictor 1, isolation index; predictor 2, sales of HCQ and CQ). Comparisons of the rates of COVID-19 cases, hospitalisations, and deaths due to COVID-19 in the peak (the inflexion point of the curve) between the states or across the macro-regions of Santa Catarina state were made using the chi-square independence test. Statistical significance was set at p < 0.05.

Results

Santa Catarina and Paraná are Brazilian states located in the southern region of Brazil. Their characteristics are listed in Table 1 .

Table 1

Geographic, demographic, and economic characteristics of Santa Catarina (SC) state and Paraná (PR) state, Brazil, 2020.

	Santa Catarina (SC)	Paraná (PR)
Population (inhabitants)	7,252,507	11,516,840
Area (Km2)	95,736	199,308
Demographic density (inhabitants/Km2)	75.8	57.8
Urban population (%)	84.7	81.4
Population aged > 60 years (%)	10.3	11.2
Per capita household monthly income (R$)	1769	1620
Municipal Human Development Index (M-HDI)	0.808	0.792
Intensive Care Units beds (beds/ten thousand inhabitants)	1.58	2.52
Latitude (state capital)	27.6oS	25.4oS

Sources: [22,24].

Data on the social isolation indices, sales of HCQ and CQ, hospitalisations due to COVID-19, and deaths due to COVID-19 according to the epidemiological weeks in 2020 in the states of Santa Catarina and Paraná are shown in Table 2 . The week that presents the inflexion point of the curve (the highest value of the series before the decline) is highlighted.

Table 2

Social isolation index, HCQ and CQ units sold, hospitalisations due to COVID-19, and deaths due to COVID-19, according to epidemiological week in the states of Santa Catarina (SC) and Paraná (PR), Brazil, 2020.

The results of confirmed COVID-19 cases are presented in Table 3 for the states of Santa Catarina and Paraná, according to epidemiological weeks and the date of onset of symptoms and date of confirmation of diagnosis. The week that presents the inflexion point of the curve (defined as the highest value of the series before the decline, representative of the peak of incidence of COVID-19) is highlighted. With reference to the dates of symptom onset, from a total of 358,997 COVID-19 cases registered in Santa Catarina, 11,901 cases (3.3%) were excluded due to inconsistencies in the registration of this variable. In Paraná, from a total of 280,084 registered COVID-19 cases, 66,602 cases (23.8%) were excluded due to inconsistencies in this registry. Regarding the dates of confirmation of diagnosis, 6937 cases (1.9%) in Santa Catarina and 624 cases (0.2%) in Paraná were excluded due to inconsistent registrations of this variable.

Table 3

COVID-19 confirmed cases, according to the date of beginning of symptoms and according to the date of diagnosis confirmation, and epidemiological week in the states of Santa Catarina (SC) and Paraná (PR), Brazil, 2020.

The evolution of the pandemic, represented by hospitalisation and death rates due to COVID-19 over the study period, together with the evolution of the social isolation indices and sale rates of HCQ and CQ, for the state of Santa Catarina, is shown in Fig. 1 a and that for the state of Paraná is shown in Fig. 1b.

Fig. 1

a) Santa Catarina state, Brazil and b) Paraná state, Brazil: Rates of hospitalisations due to COVID-19 (per million inhabitants) and rates of deaths due to COVID-19 (per million inhabitants), both registered according to the date of the beginning of symptoms, social isolation index (%) and number of chloroquine (CQ) and hydroxychloroquine (HDQ) units sold (per ten thousand inhabitants), distributed according epidemiological week of the year 2020 (week 10 begins on March 1st).

Correlation of the rise and fall in COVID-19 cases with the social isolation index and early outpatient treatment with hydroxychloroquine and chloroquine in the state of Santa Catarina, southern Brazil: a retrospective analysis.

Results of the multiple linear regression analysis with hierarchical entry of predictive variables (first, social isolation index, and second, sold HCQ and CQ units) for each outcome (hospitalisations due to COVID-19, deaths due to COVID-19, and COVID-19 cases, according to the date of onset of symptoms) in the states of Santa Catarina and Paraná are shown in Table 4 .

Table 4

Multiple linear regression with hierarchical entry of predictive variables (social isolation index and hydroxychloroquine and chloroquine units sold) for each outcome (Hospitalisations due to COVID-19, deaths due to COVID-19 and COVID-19 cases, according to the date of onset of symptoms) in the states of Santa Catarina (SC) and Paraná (PR), Brazil, 2020.

Outcome	State	Model characteristics			Predictor 1			Predictor 2
					Social isolation index			HCQ and CQ units sold
		F	R2	p	β	t	p	β	t	p
Hospitalisations	SC	29.703	0.628	<0.001*	−0.021	−0.199	0.844	0.802	7.520	<0.001*
	PR	14.989	0.451	<0.001*	−0.091	−0.691	0.494	0.665	5.029	<0.001*
Deaths	SC	29.000	0.622	<0.001*	0.062	0.581	0.565	0.813	7.562	<0.001*
	PR	12.712	0.443	<0.001*	−0.033	−0.237	0.814	0.656	4.776	<0.001*
COVID	SC	36.380	0.675	<0.001*	−0.151	−1513	0.140	0.791	7.942	<0.001*
	PR	16.905	0.483	<0.001*	−0.169	−1.320	0.196	0.651	5.078	<0.001*

HCQ, hydroxychloroquine; CQ, chloroquine; SC, Santa Catarina; PR, Paraná. *p < 0.005.

The cumulative number of hospitalisations due to COVID-19, cumulative number of deaths due to COVID-19, and cumulative COVID-19 cases (according to the date of onset of symptoms), representative data of the prevalence of COVID-19 in Santa Catarina and Paraná by epidemiological week, are shown in Table 5 . The weeks corresponding to the inflexion points of the curves of the new cases of the corresponding variable (representative of the peaks of the incidence of COVID-19) are also highlighted. There was no difference in the cumulative number of deaths due to COVID-19 per million inhabitants in the week of the inflexion point of the curve between the states of Santa Catarina and Paraná (X2 = 0.946; p = 0.331).

Table 5

Cumulative hospitalisations due to COVID-19, cumulative deaths due to COVID-19, and cumulative COVID-19 cases, according to the date of beginning of symptoms and epidemiological week, in the states of Santa Catarina (SC) and Paraná (PR), Brazil, 2020.

Table 6 shows the number of deaths due to COVID-19 per million inhabitants in each health macro-regions of the state of Santa Catarina by epidemiological week. The week that presents the inflexion point of the curve (defined as the highest number of deaths before the decline, representative of the peak of incidence of COVID-19) is highlighted. The independent chi-square test revealed a statistically significant difference in the death rates due to COVID-19 between regions of the state of Santa Catarina during the peak of the curve (X2 = 28.1, p < 0.001). In addition, the cumulative number of deaths due to COVID-19 per million inhabitants (representative of the prevalence of COVID-19) during the week of the inflexion point of the death rate curve was significantly different across the regions of Santa Catarina (X2 = 383.0; p < 0.001).

Table 6

Population, deaths due to COVID-19 per million inhabitants, according beginning of symptoms and epidemiological week (incidence), and cumulative number of deaths due to COVID-19 per million inhabitants (prevalence) during the week of the inflexion point of the death rate curve, in each region of Santa Catarina state, Brazil, 2020.

Table 7 shows the number of COVID-19-confirmed cases per hundred thousand inhabitants in each macro-region of the state of Santa Catarina (according to the date of onset of symptoms and epidemiological week, representative of the incidence of COVID-19). The week that presents the inflexion point of the curve (the peak of incidence) is highlighted. The independent chi-square test revealed a statistically significant difference in the incidence of COVID-19 across the regions of Santa Catarina during the peak of the curve before the decline began (X2 = 212.9; p < 0.001). In addition, the differences in the cumulative number of COVID-19 cases per hundred thousand inhabitants (the prevalence of COVID-19) in the week of the inflexion point of the curve of COVID-19 incidence across the regions of Santa Catarina were statistically significant (X2 = 1012.0; p < 0.001).

Table 7

Population, confirmed cases of COVID-19 per hundred thousand inhabitants, according beginning of symptoms by epidemiological week, and cumulative number of COVID-19 cases per hundred thousand inhabitants on the week of the inflexion point of the curve, in each region of Santa Catarina state, Brazil, 2020.

Table 8 shows the absolute number of HCQ and CQ units sold, the absolute number of COVID-19 cases and the proportion between these variables, according to the month during the analysed period, in Santa Catarina state.

Table 8

Absolute number of HDQ and CQ units sold in Santa Catarina state, absolute number of COVID-19 cases and the proportion between these variables according the month, in the state of Santa Catarina, Brazil, 2020.

Month	March	April	May	June	July	August	September	October
Week	10–14	15–18	19–23	24–27	28–31	32–36	37–40	41–44
HDQ/CQ (a)	29,593	142,963	104,404	107,897	285,492	237,740	139,270	144,170
COVID-19 (b)	1094	2822	13,408	38,354	84,644	55,035	26,417	49,151
Proportion (a/b)	27.1	50.7	7.8	2.8	3.4	4.3	5.3	2.9

HCQ, hydroxychloroquine; CQ, chloroquine. Sources [16,21].

Table 9 shows the absolute number of hospitalisations due to COVID-19 and its proportions to absolute number of COVID-19 confirmed cases (hospitalisation rates), according to the date of beginning of symptoms and epidemiological week, in the state of Santa Catarina, comparing two periods: the four weeks before the week of the inflexion point of the curve (the peak of incidence of hospitalisations) and the four weeks following the week of the inflexion point.

Table 9

Hospitalisation rates due to COVID-19 in the four weeks preceding the peak of incidence of hospitalisations (the inflexion point of the curve) versus hospitalisation rates in the four weeks following the peak of the curve of hospitalisations in the state of Santa Catarina, Brazil, 2020.

Week	Hospitalisations due to COVID-19	COVID-19 cases	Hospitalisation rate (%)	Total hospitalisations due to COVID-19	Total COVID-19 cases	Hospitalisation rate (%)	Chi-square
25	497	9832	5.05%
26	528	9847	5.36%
27	771	13,653	5.65%
28	1005	19,777	5.08%	2801	53,109	5.27%
Inflexion
30	1047	22,914	4.57%
31	817	18,207	4.49%
32	692	15,553	4.45%
33	602	12,708	4.74%	3158	69,382	4.55%	p < 0.001*

Sources [14,16].

Table 10 shows the absolute number of deaths due to COVID-19 and its proportions to absolute number of COVID-19 confirmed cases (lethality), according to the date of beginning of symptoms and epidemiological week, in the state of Santa Catarina, comparing two periods: the four weeks before the week of the inflexion point of the curve (the peak of incidence of deaths) and the four weeks following the week of the inflexion point.

Table 10

Death rates (lethality) due to COVID-19 in the four weeks preceding the peak of incidence of deaths (the inflexion point of the curve) versus death rates in the four weeks following the peak of the curve of hospitalisations in the state of Santa Catarina, Brasil, 2020.

Week	Deaths due to COVID-19	COVID-19 cases	Death rate (lethality,%)	Total deaths due to COVID-19	Total COVID-19 cases	Death rate (lethality,%)	Chi-square
26	134	9847	1,36%
27	213	13,653	1,56%
28	295	19,777	1,49%
29	311	23,746	1,31%	953	67,023	1.42%
Inflexion
31	248	18,207	1,36%
32	201	15,553	1,29%
33	190	12,708	1,50%
34	110	10,156	1,08%	749	56,624	1.32%	p = 0.143

Sources [14,16].

Discussion

The COVID-19 pandemic officially began in the states of Paraná and Santa Catarina when the first confirmed cases were registered in the Epidemiological Bulletin of Santa Catarina state on March 13, 2020 [16] and in that of Paraná state on March 12, 2020 [17]. Both states also initiated social isolation procedures almost simultaneously: Santa Catarina on March 18, 2020 [25] and Paraná on March 20, 2020 [26].

The incidence of hospitalisations and deaths due to COVID-19 were similar in Santa Catarina and Paraná, both in time and amplitude, during the study period (Table 2). This suggests that the evolution of the pandemic followed a similar course in both states and indicates the reliability of these statistics. In Santa Catarina and Paraná, the overburdening of health systems did not occur during the study period, with no evidence that critically ill patients were left without hospital care at any time. Registries of causes of death are usually accurate in such cases; hence, data on deaths due to COVID-19 were reliable indicators to compare the evolution of the pandemic between the two states. Furthermore, the use of the date of the appearance of first symptoms as the time reference for all outcome variables added precision to the analysis of time in this study. Therefore, the cumulative number of deaths by COVID-19 (representative of COVID-19 prevalence) in Table 5, which were strikingly similar in time and amplitude in both states, indicated that the spread of the disease occurred similarly in both states.

However, the number of registered COVID-19 cases was significantly different and was higher in Santa Catarina than in Paraná (Table 3). This could be attributed to the different underreporting rates in the two states. Identifying cases of COVID-19 in outpatient clinics may be a difficult task because many patients present only with mild symptoms. Even small differences in the levels of health surveillance and patients’ access to health services may lead to different rates of underreporting, which generates large differences in the absolute numbers of registered cases, while hospitalised cases and deaths are recorded more reliably. Therefore, the rates of COVID-19 cases should be used with caution for comparisons between the states of Santa Catarina and Paraná.

Fig. 1 reveals the associations between the variables “social isolation index” and “sales of HCQ and CQ”’ over the study period as the pandemic evolved. The social isolation index was the highest shortly after the enactment of the first social isolation measures in each state in March (the 13th epidemiological week). Since then, there has been a slow, progressive, and steady decline, regardless of the evolution of the pandemic, in both states. Multivariate analysis (Table 4) confirmed the lack of a significant association of the social isolation index with COVID-19 incidence and related hospitalisations and deaths. On the contrary, HCQ and CQ sales were strongly associated with the evolution of the pandemic. The peak of the sales of these drugs coincided with the point of occurrence of the highest COVID-19 incidence and its complications, followed by their sudden decline. Multivariate analysis revealed a statistically significant association between the three outcome variables related to the pandemic and HCQ and CQ sales in both states (Table 4).

The cause–effect relationships cannot be defined based on the statistical results of observational studies. A logical interpretation of these results must be performed carefully to generate hypotheses. The statistically significant association between HCQ and CQ sales and evolution of the pandemic found in this study could be easily explained by considering HCQ and CQ sales exclusively as an outcome of the evolution of the pandemic. As several medical practitioners performed early outpatient treatment of COVID-19 cases with HCQ or CQ, the prescriptions and sales of these drugs were expected to increase concomitantly with the increase in the number of cases and expected to decrease as these cases decrease in number. This is not the most relevant issue arising from the analysis of the results of this study. The crucial aspect is trying to understand the causes of the reduction in COVID-19 incidence, hospitalisations, and deaths due to COVID-19 observed in this study, and whether HCQ and CQ prescriptions specifically might have contributed to the control of the pandemic, as the levels of social isolation did not significantly increase even during the most intense weeks of the pandemic in the study period.

Herd immunity and transient collective immunity are concepts that could help to explain this behaviour of pandemics. The herd immunity threshold necessary to reduce the progression of an infectious disease is dependent on the transmissibility of the disease in the population, i.e., the more transmissible a disease, the greater the level of herd immunity necessary to reduce its spread. The rate of transmissibility of a disease, which can be defined by the effective reproductive number (Rt), is dependent on the social isolation practiced by a population at a certain point of time, among other factors. The level of herd immunity necessary to halt COVID-19 spread was estimated to be as low as 5.7%, where Rt = 1.06; 30%, where R t = 1.43; or 50%, where Rt = 2.00 [27]. Because the effective reproductive number decreases as the intensity of social distancing measures increases, Rt may vary heterogeneously in a territory because of different people's behaviours in many social networks. Hence, it is possible that at a certain point of time, a level of disease prevalence can be reached in limited clusters of the population that have few contacts with each other, which could temporarily halt the spread of the disease in a certain territory. This situation leads to the reduction of disease incidence, but this effect would vane easily over time as soon as social interactions within a cluster or between the different clusters of the population increases. This concept is better defined as “transient collective immunity” [28].

However, this hypothesis may be insufficient to explain the evolution of the pandemic in the state of Santa Catarina. Table 7 showed that prevalence of confirmed COVID-19 cases in Santa Catarina macro-regions, during the week of the inflexion point of incidence, ranged from 1110.3 cases/hundred thousand inhabitants (Planalto Norte/Nordeste region) to 2682.2 cases/hundred thousand inhhabitants (Foz do Itajaí region). This variability could not be attributed only to different levels of underreporting because the cumulative number of deaths due to COVID-19 per million inhabitants in each region (Table 6), the most reliable indicator to compare the spread of disease between different regions, was also significantly different across the macro-regions of Santa Catarina, at the inflexion point of incidence in each region. This strongly suggests that there were different levels of COVID-19 prevalence in the various regions of Santa Catarina at the time of the inflexion of the COVID-19 incidence curve.

The reduction in COVID-19 cases started almost simultaneously in all regions of Santa Catarina despite the different levels of pandemic spread in each of these regions. Even if social isolation indices were also significantly different across the regions of Santa Catarina, it would have been unlikely that different social isolation indices and different prevalence of COVID-19 in these regions would have occurred in such a combination that they produced the transient collective immunity necessary to halt the pandemic simultaneously in all macro-regions of the state. Hence, the number of COVID-19 cases in Santa Catarina may also have been influenced by the interference of some other factor acting in that period that may have reduced the transmission of disease.

HCQ and CQ possess effective antiviral effects in vitro against SARS-CoV-2 [29]. During the early stages of the pandemic, a study by Gautret et al. that compared 20 treated patients and 16 controls reported that HCQ was significantly effective in clearing viral nasopharyngeal carriage of SARS-CoV-2. At day 6 post-treatment, 70% of HCQ-treated patients were virologically negative compared with 12.5% in the control group [30]. Subsequently, in an observational study reported by the same group, that included 80 patients with COVID-19, 83% of them were virologically negative at day 7 post-treatment [31]. The rate of nasopharyngeal PCR positivity for SARS-COV-2 was shown to be 54% at 10–14 days post-symptoms in a recent systematic review [32]. Gautret et al. suggested that HCQ treatment could play a role in controlling the spread of the pandemic by reducing the nasopharyngeal viral load and number of days of infectivity in treated cases, thus, reducing the overall rate of transmissibility in a population in which COVID-19 cases are treated early [31]. However, reduction of SARS-CoV-2 viral load with HCQ treatment has not been identified by other researchers [33], and clinical trials have failed to demonstrate the benefits on mortality or reduced rates of hospitalisation with CQ or HCQ treatment [34]; thus, the subject remains controversial.

Studies evaluating the possible effects of HCQ and CQ on reducing the transmissibility of COVID-19 at the population level (Rt) are not available to date. It is possible that even small effect sizes of CQ or HCQ in reducing the viral load or the number of days of transmissible disease in an infected individual can reduce the Rt value at the population level if the therapy is widely applied. It is also possible that a significant effect in reducing the spread of the disease may occur with HCQ or CQ therapy if it is able to prevent the emergence of superspreaders in a region [35]. Furthermore, a patient undergoing drug treatment may comply more effectively with personalised medical orders for quarantine than with generic measures such as social distancing. All of these are possible explanations for the findings of this study.

Malaria is a rare disease in Santa Catarina. The last available record was of only two cases in 2010 [36]. Therefore, CQ and HCQ are normally used in Santa Catarina only for the treatment of autoimmune diseases, for which seasonal variation in the number of cases is not expected. Considering that the standard treatment for COVID-19 recommended by the Ministry of Health of Brazil uses six pills of CQ or HCQ [12], the number of CQ and HCQ units sold between weeks 28 and 31 (corresponding to the month of July, the peak of the pandemic in Santa Catarina), was sufficient to treat more than half of the COVID-19 cases that occurred in that month (Table 8). This is a considerable value and indicates that if an antiviral effect of CQ and HCQ is expected, it would have had an effect at the population level in Santa Catarina.

Table 9 reveals that, after the week 29 (the peak of incidence of COVID-19 cases and hospitalisations due to COVID-19), the hospitalisation rate had a small, but significant reduction. It suggests that the factor that reduced COVID-19 transmissibility also had a small effect size on reducing disease severity, resembling the actions of an antiviral pharmacological agent. The effect size on this outcome (a reduction of 0.72 percentage points in hospitalisation rate) could only be observed in clinical trials with a sample size greater than 15,000 COVID-19 cases in each group. No clinical trial using HCQ or CQ to date has analysed such large sample sizes [37].

This study had some limitations. The observational study design did not allow definitive conclusions to be drawn regarding the cause–effect relationship of the observed results. The widespread use of masks, which could have also contributed to reducing the transmissibility of COVID-19, was not assessed in this study. However, the decree that obligated masks to be used in public places in Santa Catarina was enacted on April 16, 2020 (week 16) [38], which was more than three months before the pandemic began to decline in the state; hence, its use might not have been directly related to the inflexion point of the curves. In addition, there was no information about the use of other drugs with possible antiviral effects against SARS-CoV-2. However, its use was probably much less frequent than that of HCQ and CQ during the study period. Data on HCQ and CQ sales were obtained for each month, and the weekly values were approximated by the mean. In addition, specific data about CQ and HCQ sales and social isolation indices were not available specifically for each macro-region of Santa Catarina, which could have allowed for a more precise analysis.

An argument against the hypothesis that CQ and HCQ may have contributed to the control of the pandemic would be that these medications should then have prevented the increase in cases of COVID-19 observed in July 2020. However, it can be seen in Table 8 that in the previous month (June 2020) there was the lowest number of sales of CQ and HCQ in proportion to the number of cases of COVID-19 (less than half of the cases could have been treated in the month of June). Finally, it could be possible that the severity of the pandemic in Santa Catarina might anyway decreased regardless the widespread use of CQ and HCQ. A comparison with other territory and population with similar characteristics during the same period would be helpful to assess this hypothesis. In this study, the analysis of data from the state of Paraná was intended to allow this comparison, but definitive conclusions cannot be made because CQ and HCQ were also widely used in Paraná.

The results of this study may provide hypothesis for the causes of the decrease in COVID-19 cases during the study period. The data demonstrated reliability and covered a large population over a period long enough to observe a complete wave of the pandemic in a large territory (two Brazilian states), as well as in seven regions within one of the analysed states. This was the first study to correlate early treatment with CQ or HCQ and social isolation indices with the results of the spread of the COVID-19 pandemic at a population level.

Social isolation and collective immunity may not have been the only factors that caused the decrease in the number of COVID-19 cases that occurred in Santa Catarina in the analysed period. The results of this study were consistent with the hypothesis put forward by Gautret et al. [31], i.e., they were compatible with the effect of some intervention that reduced COVID-19 transmissibility (Rt) in the study population, and CQ and HCQ were used in Santa Catarina during the study period at a sufficient scale to allow an effect at the population level.

Conclusions

In the state of Santa Catarina in southern Brazil, a sudden and almost simultaneous drop in the number of COVID-19 cases and deaths due to COVID-19 occurred across all regions of the state after weeks 28–31 of 2020. This occurred when there were significantly different levels of COVID-19 prevalence in each region, making collective immunity alone a less probable explanation for the tapering of the pandemic. The social isolation indices did not correlate with the evolution of the pandemic during this period; however, the sales of CQ and HCQ were significantly associated with it. The hypothesis that early outpatient treatment of COVID-19 with CQ or HCQ may aid in the control of the pandemic was consistent with the results of this study and warrants further research.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data statement

All data used in this research are publicly availabe, at the websites quoted in the references section.

CRediT authorship contribution statement

Fabricio Souza Neves: is responsible for the conceptualization of this work, data research, analysis of data, Writing – original draft.

Declaration of competing interest

The authors declare that there are no conflicts of interest.