Examining age, period and cohort effects in attitude change to childhood vaccinations in a representative New Zealand survey: a multiyear cohort-sequential growth modelling study

STRENGTHS AND LIMITATIONS OF THIS STUDY

  • Data were collected from a nationally representative sample of 58 654 participants from New Zealand involved in the annual, longitudinal, New Zealand Attitudes and Values Study.

  • This study used cohort-sequential latent growth modelling to disentangle age-based, cohort-based and period-based changes in attitudes towards childhood vaccination in 11 overlapping cohorts across a 7-year period.

  • Only one item was used to measure attitudes towards childhood vaccinations, ‘it is safe to vaccinate children following the standard New Zealand immunisation schedule’.

  • The data used here are from before the COVID-19 pandemic, so how the pandemic has impacted on the trajectory has not been explored.

Introduction

Vaccinations and their importance

Vaccinations are an important preventative measure in eliminating infectious diseases worldwide.1–4 Over the past 60 years, vaccinations have been responsible for the reduction of countless diseases, including human papillomavirus (HPV), smallpox, measles and hepatitis B.2 4 For example, between 2000 and 2021, the measles, mumps, rubella (MMR) vaccination successfully prevented 56 million deaths with measles deaths dropping from 761 000 in 2000 to 128 000 in 2021.5 In recent times, vaccinations have become particularly important as the primary preventative measure in decreasing the spread of COVID-19, highlighting the importance of assessing and understanding vaccination attitudes in populations worldwide.2 6 7

Childhood vaccinations are particularly important in preventing community transmission of disease and protecting the most vulnerable people in the community, for example, people who are immunocompromised.8 9 For instance, when an individual contracts the highly virulent disease of measles, they would typically spread the disease to 9 out of 10 close contacts. However, an individual who develops immunity to measles after an MMR vaccination will not become infected and will instead break the link of transmission to prevent the spread of the disease.5 Childhood vaccination schemes are designed to help protect individuals and the community as early as possible from preventable diseases, including, MMR and meningitis.2 8 9 In New Zealand (NZ), the National Childhood Immunisation Schedule protects children from birth to age 12, from a total of 13 infectious diseases, including chickenpox, MMR and HPV.10

In recent years, there has been increased concerns that antivaccination movements have reduced childhood vaccination uptake worldwide.2 11–13 Despite the rigorous safety testing and ongoing monitoring involved in approving vaccinations, there has been increasing evidence that childhood vaccination rates are not reaching the requirements for herd immunity.8 Although the vaccination rate required for herd immunity is disease-dependent, between 92-95% is recommended for childhood vaccinations.8 14 For example, to prevent measles outbreaks, community vaccination rates need to be above 95%.15 In NZ, national immunisation data show an increase in rates of fully immunised children under 5 years old from 72% in 2013 to 89% in 2019 (ie, prior to the COVID-19 pandemic).16 However, national rates of immunisation did not reach herd immunity requirements, for example, 95% for measles vaccinations.16 Furthermore, national immunisation data only examined vaccination uptake in eligible children, a proxy measure for parental vaccination attitudes. In comparison, the current study investigates attitudes in a representative sample of adults from the NZ population, thereby identifying how vaccination attitudes were changing prior to the COVID-19 pandemic across a diverse range of birth cohorts. Such knowledge is critical to understanding whether vaccine attitudes were changing prior to the COVID-19 pandemic among different segments of the population across a certain time frame. Moreover, this information is critical to develop more targeted interventions to address vaccine hesitancy within the population.

Insufficient vaccination rates can lead to harmful consequences, as demonstrated by measles outbreaks in the past.3 9 11 17 For example, in Europe, a total of 24 000 cases were reported with at least 37 vaccine-preventable deaths in 2017.11 In NZ, there have been outbreaks of previously eradicated diseases such as measles.8 18 Specifically, in 2019, a measles outbreak infected 788 children in Auckland, with 40% of those infected requiring hospitalisation.18 This could have been prevented with sufficient herd immunity from childhood vaccination.18 The reality is that despite vaccinations being available to protect more lives than ever before, more than 1.5 million children each year die from vaccine-preventable diseases worldwide.2 Vaccinations have been labelled as ‘victims of their own success,’ eliminating many infectious diseases to the point that some people no longer see a need to vaccinate due to reduced risk of exposure.3

In the current study, we examine how childhood vaccination attitudes across 11 birth cohorts changed across 7 years prior to the COVID-19 pandemic using data collected from the New Zealand Attitudes and Values Study (NZAVS). This research gives insight into whether vaccination attitudes are impacted by societal influences at large (eg, changes in norms and values) or developmental changes (ie, changes in individual characteristics across the lifespan). Findings from this study can inform the development of interventions directed at birth cohorts with more negative attitudes towards vaccinations.

Societal or age-related changes?

In economically advanced and developed countries, vaccination uptake is typically high.9 19 However, research has shown that while rates of vaccine uptake are important in identifying who is fully vaccinated, they do not determine future intentions to vaccinate.3 6 Vaccine hesitancy is a primary determinant of future vaccine uptake.3 6 In 2019, the WHO identified vaccine hesitancy as a major threat to global health, further highlighting the importance of investigating vaccination attitudes over vaccine uptake.1 8 However, previous research in Malaysia and Croatia suggests that like vaccine uptake, vaccination attitudes are generally positive.2 6

Increased dissemination of antivaccination beliefs and antivaccination movements in recent years may contribute to a decline in vaccine uptake.8 17 Changes in vaccination attitudes might be due to shifts in public opinion or through media exposure.2 6 8 12 17 20 Although research has largely focused on persuasion using television and newspapers, it has now been recognised that social media platforms have more power to generate attitude change.20 21 Internet use has increased the accessibility of personal opinions.12 17 20 This means antivaccination activists can promote and present antivaccination messaging to anyone with an internet connection without any means for correcting misinformation, for example, through social media posts.2 8 12 17 20

Furthermore, evidence shows parents are becoming increasingly dissatisfied with healthcare professionals and are turning to social media for a better understanding of vaccination safety and efficacy.8 12 17 However, the effect of social media on vaccination attitudes is not entirely detrimental. Social media can also lead to positive attitude change.20 21 For example, a meta-analysis found that social media health interventions can be used to improve vaccine attitudes and increase uptake.21 Given the widespread use of social media within the general population, especially over the past decade, there may be changes in vaccination attitudes that uniformly impact the wider population.21 In the current study, this would result in period effects, whereby all birth cohorts are influenced similarly by changing and widespread norms regarding vaccination.22 For example, in the years following the release of the falsified Wakefield article, which reported a link between the MMR vaccination and increased likelihood of developing autism, vaccination rates dropped as low as 61% in some parts of London by 2003.17 23

However, it is also possible that changes in vaccination attitudes may vary among different age groups. Previous research has identified a link between age and attitude change through multiple theoretical frameworks.22 24 For example, research largely supports the impressionable years hypothesis,24 which suggests that when a young person becomes involved in the adult world, they form the basic values, attitudes and world views that will persist throughout adulthood.22 24 This framework suggests that attitude change will be largest during early adulthood and will slow as attitudes become fixed in middle and older adulthood.22 24 In the current research, this would suggest an age effect whereby there is a change in vaccine attitudes among all birth cohorts as they grow older, independent of the period they were born.

Finally, another possibility is that individuals within the same birth cohort will display similar basic attitudes due to shared experiences during their impressionable years.22 Birth cohorts refer to a group of people who experience the same economic, social and political conditions during their lifetime (eg, people born within a 5-year period).22 In the current study, this would suggest cohort effects whereby vaccine attitudes change within cohorts born at a particular point in time, independent of the process of ageing.

Objectives of the current research

Based on the varied findings regarding age, birth cohorts and period effects on attitude change, the current study examines changes in attitudes towards childhood vaccinations over a 7-year period among 11 birth cohorts across the lifespan using a national sample of adults in NZ. As there is limited prior research examining these goals in the NZ context, our research is exploratory. If attitudes have changed over the last 7 years, a multigroup cohort sequential design will enable the differentiation between age-based, cohort-based and period-based change. The age-based model will be shown as a common developmental trend across the adult lifespan. The cohort-based model is represented as a unique change in each birth cohort, driven by historical events throughout the lives of each birth cohort. Finally, the period-based model is a similar change reflected in all birth cohorts, for example, through evolving norms which can occur through shared exposure to vaccine attitudes through the media. This research will help better understand whether societal influences or age-related variables shape changes in vaccination attitudes, which can inform future lines of research. Given the massive disruption of the COVID-19 pandemic and polarising discourse around COVID-19 vaccinations through the pandemic, the current work will focus on changes in childhood vaccination attitudes prior to the pandemic as data during the pandemic can skew the interpretation of the wider trends that preceded it. Therefore, here, we zoom in on changes in childhood vaccination between the years 2013–2019 among different cohorts of adults.

Methods

Study design and participants

Data were collected from 58 654 participants who have responded to the NZAVS over a 7-year period. The NZAVS is a national survey conducted annually from the middle of the year to the beginning of the following year. This nationwide panel study began in 2009 based on a random sample of the NZ Electoral Roll, which includes all citizens and permanent residents over 18 years of age who are eligible to vote, regardless of whether they choose to vote, barring people who had their contact details removed due to specific case-by-case concerns about privacy. Response rates over the time period covered by the current study ranged from 9.1% to 10.1% with wave-to-wave retention rates between 72% and 86% for the time frame of the study. Response rates among women are higher than men creating a response bias. However, previous research has found that trends in NZAVS data map onto results in political polling and election data based on nationally representative surveys indicating that the NZAVS reflects valid and generalisable attitudes and values at the population level.25 Further details about the survey, including response rates and generalisability, are publicly available.26

Data included in this study comprise NZ residents who responded to the NZAVS at least once over the study period. We used full-information maximum likelihood to handle missing data and include all available data because it ensures that all available information is used in the model, reducing the risk of type 1 error. Such an approach is more meaningful than arbitrarily selecting an inclusion criterion that only uses data from participants who complete a specific number of waves as this can involve systematically excluding some participants (eg, older participants who may have missed later waves due to health issues or death, or younger participants who missed earlier waves of data collection). Online supplemental table 1 provides the specific number of participants who had completed the survey for any number of times. Data on childhood vaccination attitudes were collected from wave 5 (2013) to wave 11 (2019), excluding wave 6 (2014) which did not include this specific measure due to space constraints in the national survey. Therefore, in total, there was 6 years of available data on this measure (ie, 2013, 2015, 2016, 2017, 2018 and 2019).

Supplemental material

The sample ages range from 23 to 79 years at the time of data collection. Participants were split into 11 birth cohorts, with cut-offs every 5 years between 1936 and 1990 (see online supplemental table 2). The youngest possible age within each respective birth cohort was used as an indicator of participants’ age in 2013 (wave 5, when the NZAVS first assessed vaccine attitudes). For example, the 1990–1986 birth cohort reflected a change from ages 23 to 29 (ie, 7 years). Across 7 years, each birth cohort indicated their vaccine attitudes, resulting in overlapping birth cohorts, enabling the comparison between developmental and societal changes in vaccine attitudes. For participant demographics by wave, please see online supplemental table 3. Details on the sampling procedure and sample are publicly available.26 We used the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) cohort checklist when writing our report.27

Measures

The item assessing vaccine attitudes was taken from previous research and reads as follows ‘it is safe to vaccinate children following the standard NZ immunisation schedule’.8 The item was assessed using a 7-point Likert scale (1=strongly disagree, 7=strongly agree). This item was included in the NZAVS in 2013, so the analyses that follow include all data available in the NZAVS prior to the COVID-19 pandemic.

Data analysis

The data were analysed using cohort-sequential latent growth modelling to observe changes in vaccination attitudes from the 2013–2019 period among different birth cohorts. Note that the vaccine attitude measure was not included in each wave of data collection, and thus our measurements were not evenly spaced. Specifically, vaccine attitudes were not assessed in 2014 (wave 6). A key strength of latent growth curves, relative to some other time series analyses, is that they do not assume equally spaced measurement intervals. In our case, we simply estimated the linear slope representing the rate of change in each cohort with measurement occasions coded to account for the gap introduced in 2014 (so a change of 2 annual units between 2013 and 2015, rather than 1 unit as between other sequential annual measurements).

Here, we examined three separate models: (1) an age-based model which assumes that change in vaccine attitudes follows a normative pattern across adulthood, (2) a period-based model which assumes that vaccine attitudes are changing uniformly across all birth cohorts because of shared societal norms and (3) a cohort-based model, which assumes generational differences in vaccine attitude change over time. We describe the specifications for each of these models in more detail next. All syntax for the models was made available on the NZAVS website during publication.

First, the age-based model was tested which allowed for vaccination attitudes to follow normative change throughout the adult lifespan. In this model, intercepts (i) and slopes were constrained to be equal across the 11 birth cohorts. The model trend was accounted for in estimations by including linear (s) and quadratic (q) components to allow for the possibility that change over the adult lifespan follows a curvilinear pattern. The model included a random effect for the intercept, and that random effect was constrained to be equal across cohorts. Thus, we assumed that individual differences or variability in intercepts were comparable across cohorts. Due to the complexity of the model, we did not include a random effect for slope (growth function), thus we assume that all variabilities in slopes are reflected by differences between cohorts. This assumption was needed in order for the model to converge. Results were categorised and plotted using age in years, that is, across a continuum from ages 23 to 79.

Next, a period-based model was investigated. This model is an intermediate model between the age-based and the cohort-based models. The period-based model accounts for the possibility that birth cohorts can differ in their mean levels of vaccine support but change at similar rates over time. To achieve this, the linear and quadratic slopes were constrained but intercepts were free to vary between birth cohorts (eg, one birth cohort may begin with lower levels of vaccine support than others, but all are increasing at the same rate).

Lastly, the cohort-based model was estimated, which assumed each birth cohort differed in their initial level of vaccine support and changed at different rates over time. In this model, both intercepts and linear slopes were unconstrained. However, unlike the age-based and period-based models, only linear slopes were considered in this model, not quadratic, given concerns about overfitting a model with limited data coverage at certain time points. As with the age-based model, results were plotted by age on a continuum from 23 to 79 years of age in order to compare the rates of change in vaccine attitudes between birth cohorts. To evaluate whether these data fit our models well, we used three goodness-of-fit indices and whether their values met the standard minimum thresholds outlined in the literature. These included the comparative fit index (CFI≥0.9028), the root mean square error of approximation (RMSEA≤0.0629) and the standardised root mean square residual (SRMR≤0.0830), which each provide a measure of the estimated discrepancy between the observed data and the hypothesised model. Notably, these three indices also perform better with larger sample sizes, as is the case in this study. Thus, although we report the χ2 test statistic, we do not use it to assess our models because research shows that it is unsuitable for use in analyses with large sample sizes.28

Plotted estimates for age-based and cohort-based models were overlayed against one another in order to inspect whether there is an overlap between them. If this is not the case, then it suggests changes in vaccine attitudes are attributed to cohort-based differences shaped by societal events, rather than due to normative, age-based change over time.

Patient and public involvement

None.

Results

table 1 shows the age-based model did not fit these data particularly well, (χ2(292)=11 388.37, p<0.001, CFI=0.858, RMSEA=0.084, SRMR=0.123). The estimates for best-fitting intercepts (i), linear slope (s) and quadratic slope (q) with confidence intervals (CIs) are displayed in table 2. This shows a non-significant linear slope (s=0.01, SE=0.01, p=0.224), but a significant positive quadratic growth rate (q=0.07, SE=0.00, p<0.001). As shown in figure 1 by the black line, support for vaccines declined from age 23, and then recovered and began to increase from around age 43 for the remainder of the lifespan.

Table 1

Model fit for age, period and cohort-based models for vaccine attitudes

Table 2

Parameter estimates for the ageing model for vaccine attitudes

Figure 1
Figure 1

Age-based and cohort-based models plotted against the adult lifespan. Note: Change trajectories for vaccine attitudes shown by the black line from ages 23 to 79. The lighter lines within each 5-year birth cohort panel denote longitudinal change in vaccine support across the 7 years by estimating the latent intercept (i) and linear slope (s) and overlap with subsequent birth cohorts by 2 years. Mean levels of vaccine support are shown on the y-axis across age and yearly assessments on the x-axis with 95% CIs as error bars around each point estimate. * p < .05.

The period-based model showed a modest improvement in model fit, although this still did not fit the data particularly well (χ2(282)=8547.93, p<0.001, CFI=0.894, RMSEA=0.074, SRMR=0.105) (see table 1). As shown in table 3, a general pattern emerged whereby the intercepts for vaccine support decreased with each birth cohort (ie, younger birth cohorts had more positive attitudes towards vaccines). The growth trajectory for all birth cohorts simultaneously showed a significant linear increase over time (s=0.73, SE=0.02, p<0.001), which was slightly curved (q=0.02, SE=0.01, p<0.001), for CIs see table 3.

Table 3

Parameter estimates for the period-based models for vaccine attitudes

Finally, the cohort-based model provided similar fit to these data as the period-based model (χ2(273)=8514.87, p<0.001, CFI=0.894, RMSEA=0.075, SRMR=0.105) (see table 1). As shown in table 4, each birth cohort showed significant rates of change from 2013 to 2019. As displayed by the grey lines in figure 1, although there were clear differences between birth cohorts, all showed an increase in support for vaccines over time. For CIs of intercepts and slopes, see figure 1. However, the oldest birth cohort (1940–1936) had less pronounced increases in vaccine support over time. Critically, the cohort-based estimates did not show much overlap with the age-based model, suggesting that vaccine attitudes are being driven by societal norms rather than by normative age-related changes over the lifespan. Although there are average cohort differences in vaccine attitudes, these data suggest that vaccine support is changing over time at similar rates for all birth cohorts. Overall, the period-based and cohort-based models fit the data equally well.

Table 4

Parameter estimates for the cohort models for vaccine attitudes

Discussion

This study investigated changes in vaccine attitudes across the lifespan in 11 birth cohorts over a 7-year period prior to the COVID-19 pandemic. We used multigroup cohort-sequential latent growth models on a large longitudinal and heterogeneous sample of adults from NZ. This technique allows us to better understand the distinction between normative change across the lifespan and change resulting from societal trends over time. The results show that, between 2013 and 2019, all birth cohorts displayed increasingly favourable attitudes towards childhood vaccinations. Our results show that the period- and cohort-based models better fit the data than did the age-based model. As shown in figure 1, each birth cohort showed fairly uniform increases in vaccine support over the 7-year period, suggesting a period effect whereby societal norms may be shaping more positive vaccine attitudes across generations. However, our results also show some cohort differences worth noting. For example, middle cohorts started with lower vaccine support in 2013 than their younger or older counterparts. For example, it could be that middle-aged birth cohorts were in their ‘impressionable’ years when the falsified Wakefield article23 was released in 1998 and this may have resulted in middle-aged birth cohorts having more negative attitudes towards childhood vaccinations compared with older or younger birth cohorts who were more rigid in attitude change or were born after the Wakefield article was refuted. Future research should investigate the drivers of more negative attitudes towards childhood vaccinations in middle-aged birth cohorts and more positive changes among other cohorts.

Our findings show the age-based model did not fit these data well. This suggests that age alone does not predict childhood vaccination attitudes. Furthermore, there was little difference between the cohort-based and period-based models, which fit these data better. This suggests that society may have a greater influence on vaccination attitudes, as the time period that an individual grows up in may play a more pivotal role than their age in shaping their attitudes towards childhood vaccinations. Therefore, the changes observed in the current study might be explained by a combination of societal influences, such as through media exposure, and differences in historical events throughout the individual’s lifetime. For example, research shows that the falsified Wakefield article resulted in reduced vaccination rates in London.17 23 Furthermore, it has been shown that the media played a pivotal role in the distribution of these findings, negative influencing public opinions on the MMR vaccination.23 31 Therefore, it may be that childhood vaccination attitudes in NZ could be influenced by such societal factors. The results also suggest ageing plays a partial role in predicting attitudes towards childhood vaccinations. For instance, our results show a slower rate of change in older birth cohorts, which may be explained by the impressionable years hypothesis.32

Overall, these findings are promising as they outline how vaccination attitudes were becoming increasingly positive, with all birth cohorts becoming increasingly supportive of childhood vaccinations prior to the COVID-19 pandemic. This is in line with Ministry of Health immunisation data which show a 10% increase in childhood vaccination between 2013 and 2019.16 However, older birth cohorts were found to have slower rate of change in vaccine attitudes compared with younger birth cohorts. This supports the impressionable years hypothesis, as younger birth cohorts show more flexibility in attitudes compared with middle-aged and older birth cohorts, with older birth cohorts showing the smallest changes in vaccination attitudes.22 24 This research suggests that younger birth cohorts are more likely to experience attitude changing events and may be more open to changes in attitudes.22 24 As a result, they may be more likely to be influenced by societal forces compared with older populations.

Alternatively, slower change in older birth cohorts may be a result of decreased likelihood of attitude change over time as well as changes in personality traits over time. Previous research has shown that personality tends to change across the lifespan potentially impacting on attitude change.33 For example, older adults tend to exhibit less openness to experience. Therefore, older birth cohorts may be less open to changing their opinion.33 As a result, it may be that older cohorts demonstrate less change in childhood vaccination attitudes over time compared with younger birth cohorts who exhibit more openness to experience. This is further supported by previous research suggesting that attitudes do not change throughout the lifespan and are largely formed during an individual’s formative years.32

Limitations and future research

A strength of this study is the large, longitudinal sample size, which enables the examination of vaccine attitudes across 7 years between 11 birth cohorts who together span a long stretch of adulthood (ie, ages 23–79). This study is not without limitations, including the single-item nature of the data. Although the NZAVS is a national longitudinal study, there is limited space available within these surveys to include more nuanced questions about vaccinations. Furthermore, the analytical strategy used in this research can only investigate differences based on ageing and cohort effects. It cannot determine differences observed within subpopulations, that is, ethnicity, gender, etc. Similarly, the current work cannot pinpoint exactly what societal influence may drive changes in vaccine attitudes, but it can describe what trends occur within the population. Future research should further examine the cause behind any changes in vaccination attitudes across different birth cohorts (eg, digital literacy, media access). Additionally, these results only examine changes in childhood vaccination attitudes over a 7-year period prior to the COVID-19 pandemic. While we intentionally focused on exploring trends that were emerging prepandemic to avoid the inclusion of recent waves skewing results, future work would greatly benefit from using a multiple-imputation approach to test using several years of postpandemic data if changes in childhood vaccination attitudes in a counterfactual ‘no COVID-19 pandemic’ scenario differs from observed childhood vaccination attitudes during the COVID-19 pandemic. Additionally, as increasing public attention to vaccinations during the pandemic could have contributed to polarisation effects, future research would benefit from examining attitudes towards childhood vaccinations through the pandemic using other analytical approaches such as latent class growth models to examine if polarisation among certain subpopulations has occurred.24 Another limitation of the current research is that it involves a national probability survey with higher response rates among women than men raising concerns about generalisability. Relatedly, the current work presents data from a single country (NZ) where vaccination attitudes are relatively positive. This limits generalisability of the current findings but highlights the importance of future research to examine such questions with representative samples in other nations, especially comparing these effects in a country with relatively less positive vaccine attitudes.

The practical implications of this research include informing future studies on vaccine attitudes to increase vaccine uptake. Specifically, focused interventions on younger individuals as their attitudes are more plastic than older populations. Furthermore, focusing interventions on these age groups is more likely to increase childhood vaccination rates as younger adults will likely be parents in the next 10–15 years. Therefore, increasing positive vaccination attitudes will result in an increase in vaccination rates in their children.

Conclusion

The current study reveals that childhood vaccination attitudes were becoming increasingly positive across all 11 birth cohorts between 2013 and 2019 in NZ. This is largely due to societal influences, highlighted by the close fit between the period-based and cohort-based models. Furthermore, older populations showed slower attitude change compared with younger populations, supporting the impressionable years hypothesis. The current research provides valuable insight into understanding whether there have been changes in vaccination attitudes within the larger population, and if so, whether such changes are evident in certain birth cohorts, or whether these reflect developmental changes across the life span. In NZ, at least, it appears that vaccination attitudes have become more positive in recent years and these trends have been especially pronounced in younger birth cohorts. This suggests that interventions should be aimed at younger populations to further increase vaccination attitudes and vaccine uptake among parents of children eligible for vaccination.

Data availability statement

Data are available on reasonable request. NZAVS data are hosted at the University of Auckland, New Zealand. Data cannot be made available due to ethical restrictions imposed by the University of Auckland Human Participants Ethics Committee. A deidentified dataset is available to appropriately qualified researchers on request from the corresponding author, any member of the NZAVS advisory board, or the Chair of the University of Auckland Human Participants Ethics Committee. Moreover, syntax files for the reported analyses will be made available on the NZAVS website on publication.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants and data collection for the NZAVS received ethical approval from the University of Auckland Human Participants Ethics Committee (Reference Number: UAHPEC22576). Participants gave informed consent to participate in the study before taking part.

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