Herpes zoster (HZ) vaccine coverage and confidence in Italy: a Nationwide cross-sectional study, the OBVIOUS project

This study was conducted using a cross-sectional computer-assisted web interview (CAWI). The survey was conducted by Dynata, a professional online panel provider, between April 11 and May 29, 2022, and a national sample of 10,000 respondents was recruited. The sample was defined using a stratified sampling based on proportionate allocation by first-level NUTS (Nomenclature of Territorial Units for Statistics) statistical region of residence (Northwest, Northeast, Center, South, and Islands), gender, and age group (18–24, 25–34, 35–44, 45–54, 55–64, and ≥ 65 years). Post-stratification confirmed that non-response to the survey in some strata of Italy’s adult target population had no substantial effect on the study estimates [13]. For this reason, adjustment of sampling weights was deemed as unnecessary to be performed on the target subsample of respondents for zoster vaccination (n = 1890). The data management was performed in accordance with the General Data Protection Regulation of the European Union. The survey experiment followed all requirements under Italian regulations.

The seven-sections OBVIOUS questionnaire was designed to be completed in ~ 10 minutes and aimed at investigating: demographics and living conditions, data on vaccination against HZ (the focus of this work), pneumococcus, HZ virus, rotavirus and human papillomavirus, political orientation, and attitudes toward SARS-CoV-2 vaccination, science, and alternative medicines.

The HZ section was administered to the populations that the Italian Ministry of Health guidelines identifies as vaccination targets, i.e. people aged ≥65 years and people aged ≥50 years with underlying diseases.

The questionnaire was developed based on the BeSD (Behavioural and Social Drivers of Vaccination) report on vaccine uptake [14]. The first section (sociodemographic) included18 questions: gender; date of birth; educational level; region of residence; occupation; weight; height; living conditions; ability to meet one’s needs with current income; being a parent; birth date, age and sex of the youngest child; presence of physical or mental impairment; presence of diabetes/respiratory/cardiovascular disease; where the person got the most vaccinations; where the person would have preferred to get most of the vaccinations; family and friends’ view on vaccinations.

Following this sociodemographic section, the questionnaire investigated 6 variables specific to the HZ vaccination, such as: if the respondent received the HZ vaccine; if not, whether they intended to; vaccine’s perceived safety; perceived concern about getting the disease; knowledge about the right to get the shot; and the perceived accessibility of the vaccination.

An English version of the survey questions can be found in the Supplementary Material - Questionnaire.

All variables were summarised as counts and percentages, and were stratified by first-level NUTS statistical region of residence, gender, and target group based on age and/or clinical status (ages ≥65 years vs. ages ≥50 years in conjunction with diabetes, pneumopathy or cardiopathy). Data were visualised with the aid of square charts and thematic maps with pie charts. Square charts, also called waffle charts, are a form of pie charts that use 10 × 10 grids instead of circles to represent percentages.

Multivariable multinomial logistic regression analysis was carried out to examine the drivers (determinants) of vaccine hesitancy, which was considered as a three-category nominal outcome (“I did get the vaccine” vs. “I did not get the vaccine, but I would” vs. “I do not want to get vaccinated”). In keeping with the increasing vaccination model proposed by the BeSD Expert Working Group [14], the covariates included in the regression model as potential drivers of vaccine hesitancy were the following: thoughts and beliefs about HZ infection and vaccination (perceived worry and safety concerns), social processes (friends and family’s views on vaccination, gender), and practical issues (awareness of having higher priority for vaccination, and perceived ease of access to healthcare to get the vaccine). Relevant sociodemographic determinants were also considered (age group, statistical region of residence, place of residence, degree of urbanisation, and educational attainment), as well as clinical factors that lead to a higher priority for vaccination (diabetes, pneumopathy, and cardiopathy). The effect of covariates was assessed by examining the marginal effect of changing their values on the average predicted probability of observing each outcome. The marginal effect was computed as a discrete difference in probabilities (Δ), with 95% confidence intervals (CIs) obtained with the delta method. Covariate categories occurring in < 5% of the sample were combined with adjacent lower or upper classes to improve the stability and efficiency of regression estimates. The Small–Hsiao test of independence of irrelevant alternatives (IIA) did not indicate the need for alternative model specifications in which binary logit coefficients do not converge in probability to the same values as the multinomial logit coefficients, such as the nested logit model. Lastly, in order to check for the presence of moderators, that is, covariates Z that change the effect of other independent variables X on vaccine hesitancy, we included pairwise interaction terms Z × X in the model one at a time, and tested their statistical significance with the likelihood-ratio (LR) test. To control for type I error related to multiple testing, the significance level for interactions was set at 0.01.

All analyses were conducted using Stata software, version 17 (StataCorp. 2021. Stata Statistical Software: Release 17. College Station, TX: StataCorp LP). No multi-collinearity issues were found in regression analysis, that is, the variance inflation factor was < 5 and the condition index was < 10 for each covariate.

Overall, the reported vaccine uptake in Italy was 9.6%, with a maximum uptake in the Northeast (26.9%) and a minimum in the Islands (2.9%), as illustrated in Fig. 1A. In all regions almost half of those who did not receive the vaccination reported that they would get it.

Figures 1B and 1 C illustrate the vaccine uptake stratified by gender. Males reported being vaccinated in 11.5% of cases, while females in 7.1%. Both males and females reported the highest vaccine uptake in the Northeast (33.8 and 15.0%, respectively) and the lowest in the Islands (2.3 and 3.5%).

Considering the risk conditions, as shown in Fig. 2, among the respondents ≥65 y.o.,9% reported to having received the vaccination, 46% reported not having received it but they would consider getting it, and 45% reported not having received it and would not get it. In respondents ≥50 y.o. with diabetes, 18% received the vaccination, 44% reported that they would consider getting it and 38% that would not get it. Both 10% of the respondents over 50 y.o. with pneumopathy or cardiopathy reported that they received the vaccination, 46 and 54% respectively reported that they would consider getting the shot.

As displayed by Supplemental Table 1, our investigation of vaccine-related experiences revealed that most respondents (68.3%) received most of the vaccinations in dedicated hubs; nonetheless, when asked where they would prefer to get vaccinated, the most favourite place to get the vaccine was reported to be the family doctor (36.1%), closely followed by vaccine hubs (34.5%). These results did not vary appreciably considering different regions. Regarding friends and family’s views on vaccinations, most respondents reported favourable opinions, ranging from quite favourable to very favourable in a cumulative percentage of 85.5%.

Figure 3A shows responses’ distribution regarding ease of access; overall, the access to vaccination was perceived as very/quite easy by 76.1% of our sample, with the lowest reported in the South (66.9%) and the highest in the Northeast (85.9%). As it comes to worry about the disease, displayed in Fig. 3B, the least worried region was the Northwest with 32.7% of respondents claiming not to be concerned by it. In contrast, Northeast was the most worried region, with 15.5% of respondents reporting to be very worried about the possibility of developing HZ.

Figure 3C displays perception of safety. Most Italians deemed the vaccine against shingles quite safe (66.1%); Islands were the least confident in vaccine safety, with 20.3% of respondents reporting the vaccine to be deemed from quite unsafe to very unsafe. In contrast, Italy’s Northeast was the most confident about vaccine safety, with 29.6% of respondents reporting to consider the vaccine to be very safe.

Data representation of all these matters can be found, stratified by gender and pathologic conditions, in the Supplementary material (S-Fig. A1, A2; S-Fig. B1, B2; S-Fig. C1, C2).

As shown in Table 3, being female, having high school education or less, being unworried about shingles, perceiving shingles vaccine as unsafe, having friends or relatives against vaccination, and being unaware of being a target for shingles vaccination were significantly associated with higher probability of refusing vaccination, while a significant predictor of delay in acceptance was self-reported difficulty in access to healthcare, as well as living in Insular Italy. It was also found that living in Northeastern Italy, living in rural areas, and suffering from diabetes significantly increased the probability of vaccine uptake.

The analysis of possible interaction effects across covariates revealed that the impact of not being aware of being a target for the vaccination on vaccine refusal was lower when people were quite or very worried about shingles (aware: 6.8%; unaware: 19.5%; Δ = + 12.7; 95% CI = 6.4 to 18.9) as compared to when people were not worried (aware: 43.2%; unaware: 76.1%; Δ = + 32.9; 95% CI = 21.8 to 44.0) (LR test = 24.2, p-value = 0.0022). There was also evidence of a significant interaction (LR = 25.6, p-value = 0.0012) between perceived ease of access and vaccine priority awareness, suggesting that the impact of not being aware on vaccine refusal was lower when access was perceived as very easy (aware: 29.6%; not aware: 49.1%; Δ = + 19.5; 95% CI = 8.2 to 30.9) as compared to difficult (aware: 17.3%; not aware: 45.9%; Δ = + 28.7; 95% CI = 19.3 to 38.1).

The funding source had no role in the study design, collection, analysis, and interpretation of data, in the writing of the report; and in the decision to submit the paper.

The limitations of this study are inherent to its design. In fact, being a cross-sectional study, it does not allow any cause-effect interaction between the variables analysed and only provides statistical associations. Secondly, although efforts were made to ensure the representativeness of Italy’s demographics in our sample selection, it is important to note that the recruitment process relied on an online paid survey, potentially attracting individuals seeking additional income, which may have led to an overrepresentation of individuals from lower socioeconomic backgrounds.

Finally, crucial aspects such as household income, religion, and other sensitive social characteristics were not explored in this study, thus potentially compromising the size, statistical power, and representativeness of our sample.