From intentions to practices: what drove people to get the COVID-19 vaccine? Findings from the French longitudinal socioepidemiological cohort survey

Social differentiation in vaccine intentions

In our sample (see online supplemental table 1), most respondents are women (52.1%). Nearly 40% of respondents are under 45 years old and more than a quarter are over 65. The social make-up of the sample is relatively similar to that of French society.

Based on the responses to the question on vaccination intentions in autumn 2020, the whole sample can be divided into three groups:

  • Undecided: Those who answered ‘Do not know’ represent 16% of the population (11 123 respondents). They tend to be at the bottom of the social spectrum: they are often employees (31.5% compared with 26.1% in the general population) and manual workers. Half of them (49.9%) belong to the quintile of the population with the lowest level of education, and they generally have low incomes (more than half belong to the first five income deciles).

  • Reluctant: Those who answered ‘Probably not’ or ‘Definitely not’ represent 24.7% of the population (18 274 respondents). They are mostly younger than the general population: 53% are under 44 years old (compared with 39% in the general population). They are spread across all socioprofessional categories, although employees are over-represented (30.4% compared with 26.1% in the general population). They are slightly less highly educated and have lower incomes than the general population. The descendants of non-EU immigrants are slightly over-represented.

  • Acceptant: Those who answered ‘Yes, definitely’ and ‘Yes, perhaps’ represent 59.3% of the population (51 509 respondents). The majority are men (53.8%) and tend to be older, better educated and more likely to be in higher income brackets than the rest of the population.

The intentions expressed in autumn 2020 can be compared with actual vaccination practices in summer 2021: the vaccination rate was 49.9% for the Reluctant, 67.5% for the Undecided and 82.5% for the Acceptant.

From reluctance to vaccination

The gender criterion appeared to be very important. Among the Undecided, a much larger proportion of women ultimately decided to be vaccinated than men. This gender difference remained significant when the other variables were controlled for: undecided women were 1.3 times more likely to be vaccinated than undecided men (see table 1). Among the Reluctant, there was slightly less difference in the vaccination rate between men and women, but it remained significant with all other things being equal (OR=1.2, 95% CI 1.12 to 1.29) (see table 2). In total, among women who were part of the Undecided or the Reluctant in autumn 2020, 60.5% of them later received vaccination. The women who abandoned their initial reluctance were predominantly over 55 years of age and those with the highest incomes (see online supplemental table 2).

Table 1

Probability of the Undecided in autumn 2020 being vaccinated by July 2021 (logistic regression model)

Table 2

Probability of the Reluctant in autumn 2020 being vaccinated by July 2021 (logistic regression model)

The age criterion was also a determining factor, and it was significant for all age groups. Among the Undecided, 46.4% of 18–24 year-olds were vaccinated, while this proportion was 92.5% among 75–84 year-olds. Between these two categories, the vaccination rate increased steadily in relation to age (OR=1.5 (95% CI 1.3 to 1.74) for 55–64 year-olds, OR=2.8 (95% CI 2.31 to 3.35) for 65–74 year-olds, OR=4.2 (95% CI 2.91 to 5.95) for 75–84 year-olds). However, it must be noted that this continuous increase in the vaccination rate with age did not hold for the oldest group (OR=3.0 for the over 85 year-olds). The differences in vaccination rates according to age were less pronounced among the Reluctant population, but remained substantial (OR=1.3 (95% CI 1.12 to 1.39) for 55–64 year-olds, OR=1.7 (95% CI 1.52 to 2) for 65–74 year-olds, OR=2.1 (95% CI 1.62 to 2.67) for 75–84 year-olds).

The income variable seemed to give rise to a particularly significant gradient among the Undecided: 79.2% of those who got vaccinated were in the highest decile, compared with 49.2% in the lowest decile. Between these two poles, vaccination rates increased with income (OR=0.6 (95% CI 0.52 to 0.76) in D1, OR=0.7 (95% CI 0.58 to 0.8) in D2–D3, OR=1.2 (95% CI 1 to 1.34) in D6–D7, OR=1.3 (95% CI 1.11 to 1.49) in D8–D9, OR=1.4 (95% CI 1.16 to 1.75) in D10). The trend was the same among the Reluctant population, although with smaller differences: 64.3% of those vaccinated were in the highest decile compared with 38.4% in the lowest decile. Here again, the income gradient remained highly significant when other variables were controlled for (see tables 1 and 2).

The change in trust in the government also played a determining role. According to the answers provided by the same respondents from one period to the next, 18.4% began to trust (ie, went from a No, rather not/No, not at all/Do not know answer to an expression of trust), compared with 13.9% who lost trust. Over the same period, 39.1% continued to trust, compared with 28.6% who continued to not trust. Among the Undecided, those who began to trust in the government were significantly more likely to get vaccinated (OR=1.7, 95% CI 1.46 to 1.94) than those who continued to not trust. The effect was the same among those who continued to trust.

The change in trust in scientists was even more pronounced, fluctuating at much higher levels than trust in the government. Between autumn 2020 and summer 2021, the proportion of people who said they had trust in scientists rose from 78.2% to 88.8%. Longitudinally, 70.1% of people maintained their trust in scientists, compared with only 4.9% who continued to not trust, while 18.7% began to trust, compared with 6.3% who lost trust.

Among both the Undecided and the Reluctant, those who began to trust in scientists were more likely to get vaccinated (OR=1.9 (95% CI 1.51 to 2.43) and OR=1.9 (95% CI 1.7 to 2.17)) than those who continued to not trust in scientists (see tables 1 and 2).

Finally, for both the Undecided and the Reluctant, the fact of declaring in autumn 2020 that they were afraid of catching COVID-19 increased the chances of ultimately deciding to be vaccinated.

Overall, there are certain criteria that played a determining role for both the Undecided and the Reluctant: in both cases, it was mainly women, people over 55 years of age, those with higher incomes and people afraid of catching COVID-19 who later decided to get vaccinated in spring and summer 2021. The main difference between these two groups was found in the variables related to migratory background. Among the Undecided, all those belonging to groups other than the majority group were less likely to finally get vaccinated (OR=0.5 (95% CI 0.37 to 0.8) for those from French overseas territories; OR=0.8 (95% CI 0.66 to 0.98) for descendants of EU immigrants; OR=0.6 (95% CI 0.51 to 0.78) for descendants of non-EU immigrants; OR=0.7 (95% CI 0.57 to 0.96) for EU immigrants; OR=0.6 (95% CI 0.51 to 0.78) for non-EU immigrants). Among the Reluctant, only non-EU immigrants and descendants of non-EU immigrants (OR=0.7 (95% CI 0.59 to 0.88) and OR=0.7 (95% CI 0.61 to 0.86)) were less likely than the majority group to decide to get vaccinated.

From consent to non-vaccination

The situation of people who stated in autumn 2020 that they intended to be vaccinated but were still not vaccinated by July 2021 suggested that they encountered difficulties in accessing vaccination. The data presented in online supplemental table 3 show that these people were more likely to be men (59.9%), with low levels of education (66.6% belong to the two least educated quintiles), low incomes (34.3% belong to the three lowest deciles) and more likely to be immigrants from outside the EU (10.6%) and descendants of immigrants from outside the EU (6.3%).

Supplemental material

The Acceptant (in autumn 2020) who were still not vaccinated by summer 2021 and no longer intended to be vaccinated were more often men (59.0%), with very low or no educational qualifications (47.0% belong to the least qualified quintile) and low incomes (39.9% belong to the first three deciles).

The multivariate analysis (see table 3) showed that women who were part of the Acceptant in autumn 2020 were significantly less likely than men to be unvaccinated by summer 2021 (OR=0.8, 95% CI 0.76 to 0.86). In terms of educational level, the least qualified were the most likely to remain unvaccinated (OR=1.4, 95% CI 1.26 to 1.53), and the same was true for those with the lowest incomes (OR=1.5 (95% CI 1.3 to 1.63) in D1) and for the descendants of immigrants from outside the EU (OR=1.4, 95% CI 1.21 to 1.62).

Table 3

Probability of the Acceptant in autumn 2020 not being vaccinated by July 2021 (logistic regression model)

Compared with those who continue not to trust in the scientists, those who lost trust were not significant in explaining non-vaccination (OR=1.08, 95% CI 0.88 to 1.32). In contrast, those who lost trust in government were less likely not being vaccinated (OR=1.14, 95% CI 1.04 to 1.25).

This post was originally published on https://bmjopen.bmj.com