Introduction

In France, like in many other countries, the first Covid-19 pandemic peak (March 2020—June 2020) revealed significant social inequalities in health [1–4]. The pandemic fell disproportionately on the most disadvantaged. Lower social class and racially minoritized groups were more exposed to contamination risks [5], and more often contracted severe forms of the virus [6], along with older individuals [7]. Thus, there is a need to analyze the medical care that these groups received when contracting the virus.

In France, during that period, access to care was deeply impacted. Indeed, the French system mostly relied on a containment strategy of the disease and social distancing in order to reduce new cases and delay the influx of acute patients in hospital. The most radical measure took place from March 17^th to May 10^th, when a strict lockdown was implemented. Patients with Covid-19 symptoms were appealed to first contact a GP by phone, except for severe symptoms, in which case the recommendation was to call the French Emergency Medical Assistance Service. Furthermore, the French Health System was reorganized, resulting in a drastic health care rationing [8]. Health professionals activities decreased during that time period by 60% for specialists and 30% for GPs and midwives [9]. Moreover, doctors had to adapt their practices. The use of telemedicine was amplified: in April 2020, 27% of medical consultations were done with telemedicine, compared to less than 1% before the outbreak [10].

Access to care was not the only possible barrier to receiving medical attention when contracting Covid-19-like symptoms. Indeed, marked social inequalities in health care services utilization existed in France prior to the pandemic and had been well documented [11–13]. For the same health needs, ethno-racial minorities and people with lower education and income less frequently used health care than the rest of the population [14]. These inequalities were even more pronounced for care provided by specialists. A lower or later use in health care services could have contributed to worsen the impact of contracting the coronavirus, even more so for the most vulnerable.

There is a need to measure whether the French health system has been able to respond to this unprecedented health situation. There is also a need to clarify if overrepresented social groups in contaminations were also overrepresented in health care services utilization, when experiencing Covid-19-like symptoms. To our knowledge, this has not been done in the literature. However, two studies reported health care seeking due to Covid-19-like symptoms among older people, one in the US (n = 697) [15, 16] and one in Brazil (n = 568) [16]. While two different rates of people who sought care were found in these studies, 74.0% for the first and 33.6% for the second, neither of the studies found differences, according to race or to social factors, in seeking health care.

Using a large French population-based random survey, the aim of our study was to provide new insights on this topic by exploring the use of health care services of people with Covid-19-like symptoms.

Participants and methods

Statistical analyses

Use of health care services was first described according to the respondents’ demographic, socio-economic (Table 1), and health characteristics (Table 2). In order to test the association between use of health care services and social factors, adjusted odds ratios (ORs) on all variables presented in Tables 1 and 2. were assessed by multinomial logistic regression. All analyses were weighted using a Heckman model to take into account the effect of the response mode on the reporting of Covid-19-like symptoms. The threshold value for p was <0.05. Statistical analyses were performed using SAS 9.4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Distribution of the use of health care services by socio-demographic characteristics.

https://doi.org/10.1371/journal.pone.0279538.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Distribution of the use of health care services by health characteristics.

https://doi.org/10.1371/journal.pone.0279538.t002

Results

In our sample, 37.8% of those who reported cough, fever, dyspnea, anosmia, ageusia and/or dysgeusia consulted a doctor for these symptoms; 32.1% outside hospital and 5.7% in hospital (Table 1). Women used health care services more than men with stronger differences for out-of-hospital use consultation(s) (34.7% versus 28.8%). There was a clear positive gradient between age and consultation(s) rates: the older individuals were, the more they consulted at least once, whether it be in-hospital consultation(s) (2.5% of the 18–24 years versus 15.6% of the 75 years or more) or out-of-hospital consultation(s) (24.4% versus 37.8% outside hospital). Differences in consultation(s) rates were also found according to the ethno-racial statuts. FOD or descendants of FOD natives and first-generation immigrants more often consulted doctors than the mainstream population. The highest in-hospital consultation(s) rate was found among racially minoritized first-generation immigrants (10.4% versus 5.2% of the mainstream population). Health care utilization was also strongly associated with social position. Senior executive professionals consulted doctors less often than manual workers (34.3% and 38.4% respectively), with marked differences regarding in-hospital consultation(s) (3.8% versus 7.1%). A similar trend was found regarding incomes with a lower in-hospital consultation(s) rate among the richest, compared to the poorest (4.0% versus 7.2%). Finally, there was a distinct negative gradient between in-hospital consultation(s) and education: the more educated the respondents were, the less they used in-hospital health care services (from 11.1% for those without any diploma to 3.5% for those with the highest education level).

When it comes to health variables, 62.4% of the respondents attributed the Covid-19-like symptoms to an infection to the coronavirus and the symptoms lasted for over than a week for 61.7% of the respondents (Table 2). Respondents who suffered from a chronic disease consulted a doctor more often than the rest of the population. The highest rates were observed among those who suffered from cardiac diseases (50.5% versus 37.8%). More generally, respondents who perceived their health as bad consulted more often (53.9% versus 37.0%), especially in hospitals (15.3% versus 5.2%).

Results from the multinomial analysis comparing use of health care services (in-hospital consultation(s), out-of-hospital consultation(s) versus no consultation) among those with Covid-19-like symptoms are presented in Table 3. After adjustments on health variables, characteristics of Covid-19-like symptoms and variables related to consultation(s) accessibility and needs of health care, results confirmed that women were more likely than men to consult a doctor, but only for out-of-hospital consultation(s) (OR, 95% CI: 1.25 (1.13–1.37)). Positive association between age and medical consultation(s) was found and was the strongest regarding in-hospital care (compared to the 18–24, up to 9.51 (5.02–18.0) for 75 and older for in-hospital care and 2.58 (1.73–3.84) for out-of-hospital care). Racially minoritized first-generation immigrants consulted doctors more often than the mainstream population, but for in-hospital care (1.61 (1.09–2.36)). While no association was found between social class and health care use, results showed a negative association with level of education: the more educated respondents were, the less consulted doctors, with larger gaps for in-hospital consultation(s). Compared to respondents with the highest education level, respondents with no diploma or a primary education level used in- and out-of-hospital care more often (respectively 1.34 (1.08–1.67) and 2.20 (1.44–3.38)). Similar associations were found with standard of living, with a greater use of consultation(s) rates among middle deciles (respectively 1.11 (1.00–1.24) and 1.38 (1.10–1.74), compared to the richest 30%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Factors associated with use of health care services (multinomial regression, reference = no consultation).

https://doi.org/10.1371/journal.pone.0279538.t003

Discussion

To our knowledge, this study is the first to describe use of health care services among people aged 18 and over with Covid-19-like symptoms during the first wave of the epidemic. Regardless of their health status and the severity of their Covid-19 related symptoms, we found that women, older adults, disadvantaged social groups and racialized first-generation immigrants were significantly more likely to use health care services, in particular in-hospital care, than the rest of the population.

The health care utilization patterns observed during our study duplicated those before Covid-19. They may also have been reinforced by the Covid-19 peak context. Indeed, women being both recipients and providers of healthcare [19], they are used to visiting GPs more often than men [20]. Pre-Covid-19 evidence shows that older adults consulted doctors (in- and out-of-hospital) more often than younger adults [21]. Older adults were also particularly targeted by prevention campaigns during the first pandemic peak, which contributes to explaining why for the same symptom severity, they used health care services more often. The most socially disadvantaged are known to use GP services and to consult doctors in hospitals more often than the rest of the population. An explanation lies in the nature of the French Health system that guarantees an extensive coverage of health expenditures. Furthermore, in a study on over 3,000 GPs, 1 GP out of 4 reported initiatives to contact patients during lockdown with a strategy including social criteria [22]. Similar pre-Covid-19 patterns were also found according to ethno-racial status [23–25]. Immigrants’ higher in-hospital consultations rate could be partly explained by a lower use of telemedicine. A study among NYC patients showed that difficulties in the utilization of digital healthcare among the ethnic minorities continued to be observed during the early phase of the pandemic [26]. Contrary to the most disadvantaged and ethno-racial minorities, the most advantaged and the mainstream population refer to specialists to a greater account. It is noteworthy that during the first Covid-19 peak, specialists of the disease did not yet exist. As a matter of fact, nearly all the options to receive medical care were in-hospital and GPs could be consulted for expert advice.

Very few social differences in health care utilization during the first Covid-19 peak when presenting Covid-19-like symptoms were found. Only two studies looked into rates of health care seeking among older people with Covid-19-like symptoms by race in the US [15] and social factors in the US and in Brazil [15, 16]. Although the studies did not take place during lockdown, they were still conducted at a time of a peak of the pandemic and great disruption to the Health systems. In the Brazilian study (n = 568), authors found that 33.6% of people aged 50 or over sought care for Covid-19 related symptoms. This rate reached 74.0% in the US study (n = 697), where the average age was 76.8 years. In our study, 45.0% of people aged 50 and over had at least one medical consultation and the rate rose to 53.4% for those aged 75 and over. In both studies, no health inequalities regarding health care seeking was found, according to race (US study), sex, age (Brazilian study), income and education (both studies). When we restricted our analyses to people aged 50 and over (n = 4177, S1 Table), differences still persisted according to social factors: the older and the less educated respondents were, the more they used in-hospital health care. Conflicting results with the two previous studies may result from several factors. First, we studied health use of health care services and not health care seeking. Although in our study, among people who did not consult a doctor, only 1.0% did not due to barriers to health care access (no nearby doctor, long delays to get doctor’s appointment, financial costs). Moreover, sensitive analyses in the Brazilian study suggest that barriers to health care access were not associated with care seeking. Due to fewer included individuals, previous studies may suffer from lack of statistical power.

The social differences that our study found in health care services utilization might be in some measure explained by other factors, such as the fear of the disease. The poorest and the least educated have been shown to have higher levels of Covid-19 fear [27–29], which could potentially lead them to consult a doctor or go to the hospital more often than their counterparts when presenting Covid-19-like symptoms. Compared to the mainstream population, racialized minorities accumulated more exposure factors during the pandemic peak [5], and lived more often in overcrowded households [5], which could have fueled the feeling of fear of the virus.

Our study shows limitations. First, as any national population-based survey, the study fails to capture highly vulnerable groups such as undocumented migrants and homeless people, who are particularly affected by the pandemic [30]. Additionally, it is possible that the most severe cases have been under-represented in our study, which may relate to the most socially disadvantaged and ethno-racial minorities more [1]. Secondly, participants were questioned on their use of health care services for at least one of 12 items list of symptoms suggestive of Covid-19 while we have retained only five. It is therefore possible that some participants reported use of health care services for other symptom(s) than those selected.

Several sensitivity analyses were done. We excluded individuals whose symptoms appeared in the week preceding their response to the questionnaire, as they had a shorter period of time to visit a doctor than others. We then adjusted our final model on the fear of contracting Covid-19 when consulting a doctor. In both cases results and conclusions remained unchanged. As we studied health care access, people who sought care and did not have any medical consultation could not be identified. However, in another question, we asked participants what they first did at the onset of the first symptom: among those who did not use health care services during the whole study period, less than 2% unsuccessfully sought care. Therefore, one might think that a very low percent of people who did use health services at the onset of the symptom unsuccessfully sought care after. Finally, in our analyses, the region variable was included as a control variable even though it is a contextual variable. We therefore calculated clustered sandwich estimators, which are used to adjust inference when errors are correlated within clusters (e.g. region). Our results were very similar, to the nearest hundredth, to those we obtained when we simply adjust on region. This means that biases resulting from not taking the region variable as a cluster seem to be minors.

In conclusion, our results seem to reflect pre-existing social differences in health care utilization patterns. Those patterns might have been strengthened by the context of the Covid-19 peak, and the resulting fear of the virus. In fact, social groups who were the most at risk were also those who have consulted the most. As a consequence, the potential impact of social inequalities in exposure and infection to the Covid-19 [5] seemed not to have been widened by the use of a free health care system.

Supporting information

Acknowledgments

The authors warmly thank all the volunteers of the EpiCov cohort; the DREES and INSEE teams; the staff of IPSOS, Inserm Santé Publique team, and Frédéric Robergeau.

The EpiCoV study group: Nathalie Bajos (co-principal investigator), Josiane Warszawski (co-principal investigator), Guillaume Bagein, Emilie Counil, Anne Gosselin, Florence Jusot, Nathalie Lydie, Claude Martin, Laurence Meyer, Ariane Pailhé, Delphine Rahib, Philippe Raynaud, Alexandra Rouquette, Patrick Sillard, Alexis Spire.