Changes in general practitioners’ attitudes toward COVID-19 vaccination after first interim results: a longitudinal approach in France

ABSTRACT

We assessed whether the a priori acceptance by French general practitioners (GPs) of COVID-19 vaccines changed after the announcements about them in November 2020. In all, between two surveys in October-November and in November-December 2020, acceptance of COVID-19 vaccines increased among 16.9% of GPs and decreased among 23.0%. Among those with high acceptance in October-November (52.5%), 11.6% became hesitant-reluctant in November-December; in those with initial hesitancy-reluctance (24.6%), 15.2% showed high acceptance. Deteriorating acceptance was significantly associated with GPs’ distrust in the Ministry of Health and a priori concerns about the safety of vaccines developed during an epidemic; the reverse was found for improving acceptance. In addition, better acceptance was more likely among GPs who perceived the medical severity of COVID-19 to be high and was less common among women. During a severe pandemic, GPs’ trust in health institutions and perception of safety issues remain important predictors of their attitudes toward new vaccines. Vigilance is needed regarding health professionals’ reactions to events that may cast doubt on the safety or efficacy of certain COVID-19 vaccines. Personalized approaches should be considered and tested to address their concerns as the situation and knowledge evolve.

Introduction

GPs are the cornerstone of general population immunization in France and have been entitled to vaccinate against COVID-19 since March 2021. However, some of them find – or found in the past – the usefulness or safety of certain vaccines (e.g. against seasonal and A/H1N1 pandemic influenza) uncertain, and this negatively affects their willingness to recommend these vaccines to their patients.^1,2 Moreover, in a survey we conducted from October 6 to November 16, 2020, among GPs, a quarter were hesitant or reluctant about future COVID-19 vaccines, either to vaccinate themselves or to recommend them to their patients.³ The main driver of these attitudes was their distrust in the safety of vaccines developed rapidly during a pandemic. These observations took place when the safety and efficacy profiles of the new COVID-19 vaccines were not yet known. These profiles became less hypothetical as several pharmaceutical companies developing COVID-19 vaccines successively announced their preliminary results starting on November 9, phase 3 trial results were published, and the first marketing authorizations granted (Figure 1). Given the high level of hesitancy toward these vaccines in the French general population, which GPs must address during the vaccination campaign,^4,5 as well as among health care workers,⁶ it is important to understand how GPs reacted in this unprecedented context, for this response may provide insight into the effectiveness of the health authorities’ information strategy toward this profession. This article sought to analyze how GPs’ attitudes toward COVID-19 vaccines changed after November 9 and the factors associated with these changes.

Figure 1.

Timeline of Covid-19 vaccine milestones and survey dates

^aInterim results transmitted by the pharmaceutical company. ^bAstraZeneca. ^cMedicines and Healthcare products Regulatory Agency. ^dFood and Drug Administration. ^eEuropean Medicines Agency. ^fOnly 30 GPs among the 800 on the panel responded to the first survey between 11–15 November.

Methods

We conducted a second cross-sectional questionnaire-based survey from November 24 to December 27, 2020, with the same methodology as the survey mentioned above (which took place from October 6 to November 16): the present article focuses on participants to the two surveys at the two time-periods. Both surveys were implemented among a national representative panel of 2,755 non-salaried GPs, set up in 2018 by random selection of GPs from an exhaustive database of health professionals (French national directory of health professionals).³ Panel members were invited to participate online and received at least five reminders if they had not completed the survey.³ To be included, GPs had to be in private practice and not practice complementary and alternative medicine exclusively.

The questionnaires concerned “future” COVID-19 vaccines, viewed from the survey period, before any such vaccine was authorized or available in France. Prior to conducting the previous and current study, we pilot-tested them for clarity, length, and face validity with 50 GPs. The two principal questions focused on participants’ willingness: (i) to be personally vaccinated, and (ii) to recommend the vaccines to their patients. Answers used a five-point scale from “no, certainly not” (score = 0) and “no, probably not” (score = 1) to “yes, probably” (score = 2), and “yes, certainly” (score = 3), with a “do not know” option.

In addition to questions on the acceptance of future COVID-19 vaccines, the questionnaire assessed the participants’ trust in the authorities, their perception of the severity of the COVID-19 pandemic, the safety of vaccines developed in the event of an epidemic, their history of seasonal influenza vaccination as well as their demographic and occupational characteristics. In each survey, we ensured that each respondent could complete the questionnaires only once.

Applying a previously published approach,³ we constructed a score of presumptive acceptance of future COVID-19 vaccines based on responses to these two questions (range: 0–6). We used the resulting score to categorize participants by their degree of “COVID-19 vaccine acceptance”: “high acceptance” (score > 4), “moderate acceptance” (score = 4) and “hesitancy or reluctance” (score < 4 or answers “do not know” to at least one question). Hesitancy and reluctance were grouped together to obtain sufficient numbers in this category. It is acknowledged that degrees of vaccine hesitancy can follow a continuum between acceptance of all vaccines and their complete refusal with no doubts, with vaccine hesitant individuals the heterogeneous group between these two extremes.⁷

We weighted the samples for age, gender, region, and workload. Workload was based on each GP’s number of consultations and patient visits in 2017, obtained from the French National Health Insurance Fund.

Multiple logistic regression models analyzed the factors – measured in October-November 2020 – associated with marked changes in GPs’ COVID-19 vaccine acceptance level between the two surveys (change from high acceptance to hesitation-reluctance for Model 1 and vice versa for Model 2). We targeted this type of change because it is less expected and more radical than changes in attitude to or from the intermediate category (moderate acceptance) of the acceptance variable (Figure 2).

Figure 2.

Changes in acceptance of future COVID-19 vaccines (N = 805)^a

^aAll figures and percentages are weighted (for gender, age, region, workload). Reading note: Of the 198 general practitioners (25% of the 805 participants) who were hesitant or reluctant in the first survey in October-November 2020, 58% remained so in the second survey in November-December 2020, 27% moved into the Moderate Acceptance category, and 15% into the High Acceptance category.

To test whether potential differences between panel participants at inclusion and those participating in the two surveys at the end of 2020 might bias the results of the regression analysis studying factors associated with changes in GPs’ attitudes toward the COVID-19 vaccines, we implemented a probit model with sample selection (Supplemental material S1).^8,9 All analyses used two-sided p-values, defined statistical significance as p < .10, and were performed with Stata 14.

Results

Overall, 805 GPs (29%) participated in both surveys (see Supplemental material S2). In October-November, high acceptance of future COVID-19 vaccines was observed in 52.5% (95% confidence interval (CI) 47.0, 57.9), moderate acceptance in 22.9% (95%CI 18.8, 27.7), and hesitancy-reluctance in 24.6% (95%CI 20.4, 29.2) (Figure 2). From October through December, in GPs with initial high acceptance, this remained stable for 66.6%, became moderate for 21.8%, and evolved to hesitancy-reluctance for 11.6%; in those with initial moderate acceptance, it became high in 28.1%, remained stable in 48.1%, and deteriorated to hesitancy-reluctance in 23.8%. Finally, of the initially hesitant-reluctant GPs, attitudes did not change for 57.6%, moderate acceptance appeared for 27.3%, and high acceptance for 15.2%. At the same time, physicians’ perceptions that new vaccines developed during an epidemic may not be safe rose for 19.4%, remained stable for 57.0%, and decreased for 23.6%. The frequency of marked changes (from hesitancy-reluctance to high acceptance and vice versa) was not significantly different between the period before the first authorization and scientific publication about these vaccines (before December 2, 2020, Figure 1, Table 1) and the following period.

Table 1.

Frequency of marked changes in acceptance of future COVID-19 vaccines by period, weighted data, N = 474^a.

	Before 02/12/2020		02/12/2020 onwards		Total
	(N = 298)		(N = 176)		(N = 474)
	N	Col. %	N	Col. %	N	%
Decrease from high acceptance to hesitancy/reluctance	(N = 212)		(N = 118)		(N = 330)
Stayed high	179	84.53	102	86.10	281	85.09
Hesitancy/reluctance	33	15.47	16	13.90	49	14.91
P-value:	0.81
Increase from hesitancy/reluctance to high acceptance	(N = 86)		(N = 58)		(N = 144)
Stayed low	71	82.59	43	74.25	114	79.23
High acceptance	15	17.41	15	25.75	30	20.77
P-value:	0.45

^aThe two periods were defined as follows: survey period before December 2, 2020, corresponding to the availability of the pharmaceutical companies’ press releases about the preliminary results for their vaccines; survey period starting December 2, 2020, corresponding to the first marketing authorizations and peer-reviewed scientific publications about the efficacy and safety of the first vaccines (Pfizer, Moderna and AstraZeneca).

Regression model 1 showed that the probability of a transformation to hesitancy-reluctance from initial high acceptance was significantly associated with reporting mistrust that the Ministry of Health would adequately control the safety of new vaccines, and concerns about the safety of vaccines developed during an epidemic in the October-November 2020 survey (Table 2). Regression model 2 found that the probability of high acceptance after initial hesitancy-reluctance was significantly associated with male gender and reporting in October-November, initial trust in the safety of vaccines developed during an epidemic, and the perception that the medical severity of the pandemic was moderate to high (Table 2). The probit model found no issue of selection bias in the results of models 1 and 2 (Supplemental material S1).

Table 2.

Factors associated with marked changes in the acceptance of future COVID-19 vaccines between October-November (survey 1) and November-December (survey 2), logistic regressions, weighted data

	Model 1. Decrease from high acceptance to hesitancy/reluctance (ref. stayed high)		Model 2. Increase from hesitancy/reluctance to high acceptance (ref. stayed low)
	N: 330		N: 144
	aORa	[90% CI]b	aOR	[90% CI]
Characteristics
Gender (ref. Male)
Female	1.01	[0.42,2.43]	0.25**	[0.09,0.68]
Age (ref. >50 years)
50 to 59	0.38*	[0.15,0.98]	2.49	[0.84,7.40]
≥ 60	0.41	[0.14,1.26]	0.54	[0.12,2.46]
Occasional practice of complementary and alternative medicine (ref. No)
Yes	1.14	[0.30,4.34]	1.94	[0.59,6.39]
Personal vaccination
Vaccinated against seasonal influenza for the winter 2019/20 season (ref. No)
Yes	0.37	[0.10,1.38]	0.60	[0.17,2.15]
Trust in the Ministry of health, in October-November 2020
I trust the ministry of health to ensure the safety of vaccines in general (ref. Disagree/don’t know)
Agree	0.10****	[0.04,0.23]	1.08	[0.41,2.84]
Perceived risks, in October-November 2020
Perceived medical severity, for the population, of the Covid-19 epidemic (ref. Low)
High/moderate	1.00	[0.38,2.63]	7.56***	[2.20,26.03]
The safety of a vaccine developed in an emergency, during an epidemic, cannot be considered guaranteed (ref. Disagree)
Agree	3.23**	[1.45,7.20]	0.27*	[0.09,0.81]
Don’t know	3.57	[0.83,15.35]	0.16*	[0.03,0.80]

^aAdjusted Odds Ratios; ^b 90% Confidence Interval.

p-values: * p < .10; ** p < .05; *** p < .01; **** p < .001

Due to the limited number of observations in the models, we used 90% confidence intervals.

Discussion

This study is the first to show trends in individual attitudes of health professionals in December 2020, in the context of a vaccine infodemic, marked by a multiplication of announcements, publications, and decisions in favor of effective and safe COVID-19 vaccines (Figure 1). While a small majority of GPs did not change their attitudes at that time, they evolved for the others in opposite directions, depending on their a priori attitudes toward these vaccines and their level of confidence in the health authorities. We found no indication of a majority shift in favor of vaccines. These surprising results must be put into the context of the time.

The first vaccine announcements by the pharmaceutical industry were initially received with circumspection by certain French scientific or medical personalities because they were not peer-reviewed; these prominent individuals were also concerned about the “radically new” technology of mRNA-based vaccines, despite its years of development. Renowned French medical experts (some of whom later retracted their statements) and the National College of General Medicine Teachers¹⁰ called for caution toward these vaccines due to the lack of sufficient follow-up time in phase 3 and 4 trials; they stated that they would wait before vaccinating themselves and emphasized the need to wait for scientific publications. This reaction is understandable: after the decade-long Mediator scandal, French healthcare professionals have a low level of trust in pharmaceutical companies.¹¹

It is therefore quite logical that the attitudes of the GPs who had the least confidence in the health authorities and the most distrust a priori of rapidly developed vaccines evolved negatively (Table 2). They probably distrusted the statements made by vaccine manufacturers and awaited for authorizations from the French drug agency. For these doctors, the multiplication of sources of information and their overexposure to information about vaccines could have also promoted a deterioration in their attitudes. Excessive exposure to information from multiple sources during the pandemic was shown to be a major difficulty and stressor for GPs.¹² A significant proportion of French GPs have relatively little confidence in the health authorities – close to 40% in 2020.¹³ The current official information channels do not appear to have had a sufficiently positive and reassuring impact on them. This result should lead to a review of the information strategy for these doctors, based, for example, on a more detailed collection of their information needs and of the reasons for their mistrust or even their grievances. Involvement of GPs in the design of the vaccination campaign and the information strategy of health care professionals has been attempted, as a lesson learned from the previous A/H1N1 pandemic in 2009,¹⁴ but perhaps still too timidly. It is nonetheless a difficult task given the heterogeneity of the profession, particularly in terms of practice conditions (e.g., individual or group practices) and representation by various unions and professional associations, which are not necessarily propitious to trust in health authorities.

In contrast, GPs who had more confidence in the safety of the vaccines were significantly more likely than others to report a positive change in their attitudes toward COVID-19 vaccines when initially hesitant. This group was more likely to be male, which is consistent with cross-sectional studies indicating that women’s attitudes toward COVID vaccines are more likely to be unfavorable, both among health professionals and the general population.¹⁵ This positive time trend was also more frequent among doctors whose perception of the severity of the COVID-19 epidemic had not moved in the direction of trivializing it by the end of 2020. Between April and December 2020, the proportion of GPs perceiving the epidemic’s severity as low rose from 14% to 30% and the proportion perceiving this severity as high fell from 39% to 19% in France,¹⁶ results that contrast with much higher levels of risk perception among hospital staff.¹⁷

Strengths and limitations

An existing panel of GPs offered the opportunity to analyze their individual changes in COVID-19 vaccine acceptance. Although participation rates were relatively high for online surveys among such hard-to-reach populations, the number of GPs participating in both was not. We were able, however, to weight our data for variables associated with participation and could verify selection bias does not appear to be an issue in our data (Appendix). Causal inferences cannot be drawn from such an observational study.

Conclusion

In the current pandemic, media dissemination of reassuring and positive information about the efficacy and safety of the new vaccines did not eliminate hesitancy toward them, even among GPs. The instability of the attitudes of a fraction of GPs in December 2020 underlines that these professionals remain sensitive to new information and/or controversy concerning the efficacy and safety of COVID-19 vaccines, such as those concerning the AstraZeneca vaccine during the first trimester of 2021 (occurrence of thromboembolic events¹⁸ and different management of these events from one country to another in Europe, including recommendations on the target groups to which this vaccine should be limited).

Monitoring healthcare professionals’ attitudes and behaviors toward COVID-19 vaccines remains essential. Longitudinal approaches would make it possible to observe changes at the individual level and to study and quantify the effect of the factors directly influencing these changes, which is required to design effective interventions.¹⁹ These approaches should also be complemented or even preceded by qualitative studies to understand in more depth what drives these professionals’ attitudes and behavior and their specific needs in terms of information.

Personalized approaches are likely to be needed to address the concerns of each professional as the situation and knowledge evolve. This is a considerable challenge given the scale of the task. Interactive tools that are easily adaptable to new vaccine information and counter-arguments and can be widely disseminated are an avenue worth exploring.²⁰

Funding Statement

Direction de la Recherche, des Etudes, de l’Evaluation et des Statistiques (DREES)/Ministère des solidarités et de la santé [grant 2102173353]; Agence Nationale de la Recherche (ANR) [ANR-20-COV8-0009] in the frame of the call for projects 2020 « Recherche-Action Covid-19 ».

Supplemental data

Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/21645515.2021.1943990

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Ethics statement

The ethics board of the Conseil national de l’information statistique (France, CNIS, avis n°114/H030) approved the study.