Effect of CARD (Comfort-Ask-Relax-Distract) on acceptance of school-based vaccinations: A controlled before and after study

ABSTRACT

Injection-related pain and fear are common in children undergoing school-based vaccinations and contribute to vaccine refusal. The CARD system (Comfort Ask Relax Distract) includes interventions that reduce pain and fear. This pragmatic study evaluated CARD’s impact on school program-related vaccine uptake using a controlled before and after study design. Altogether, 67 Public and Catholic schools receiving vaccination services from a public health unit in Ontario, Canada, were included. Schools were ranked by vaccination uptake in grade 7 students in 2022–2023. The bottom 29 schools were allocated to CARD (intervention) and the remainder to control for 2023–2024 school vaccinations. In CARD schools, nurses educated students at school and information about CARD was added to vaccine correspondence for parents. On vaccination day, nurses applied interventions designed to improve coping. Vaccination coverage outcomes were assessed for 3 targeted vaccines – human papillomavirus (HPV), hepatitis B (HB), and meningococcal conjugate-ACYW (MCV). Baseline (2022–2023) vaccine coverage was lower (p < .01) in CARD-assigned schools (n = 1478) [vs. controls (n = 1729)] for all vaccines. During the study period (2023–2024), coverage increased (p < .001) within the CARD group (n = 1403) for HPV (+9.8%), HB (+8.1%), and MCV (+5.3%). Coverage remained the same or decreased within the control group (n = 1877): HPV (−1.1%; p = .50), HB (−4.4%; p = .004), and MCV (−2.7%; p = .03). In 2023–2024, vaccination uptake did not significantly differ between groups. Overall uptake in all schools was higher in 2023–2024 vs. 2022–2023 for HPV (+4.0%; p = .001). This study demonstrates effectiveness of CARD for increasing vaccination uptake. Evaluation of impact in other geographical regions is warranted.

Introduction

Injection-related pain, fear and associated immunization stress-related responses are common in children undergoing vaccinations at school and can contribute to vaccine hesitancy.^1–4 In a recent systematic review, we demonstrated that the prevalence of childhood vaccine refusals related to concerns about pain and fear is 8%.⁵ Despite the large body of literature on interventions to reduce these stress related responses,⁶ they are not systematically integrated into school vaccination delivery in Canada.⁷

The CARD system (Comfort Ask Relax Distract) is an evidence-based vaccine delivery framework that was developed to address this knowledge-to-care gap.⁸ Vaccine recipients select coping options from the different letters of the CARD acronym (C-A-R-D) to reduce negative symptoms (e.g., pain, fear) and to improve their vaccination experience. Providers follow the CARD 4E model (Education, Environment, Engagement, Evaluation), which includes supporting vaccine recipient coping choices as well as using other evidence-based pain interventions (e.g., reducing fear cues; see also: www.cardsystem.ca).

In three prior cluster trials, we demonstrated that CARD reduced pain, fear and related immunization stress-related responses in children undergoing vaccinations at school.^9–11 The impact on vaccination coverage, however, could not be adequately evaluated. One of the trials was too small,⁹ and the other two were interrupted by the COVID-19 pandemic,^10,11 resulting in only one of the two regularly scheduled school vaccination clinics being conducted. As two of the three targeted vaccines require two doses for series completion, this prevented an accurate estimate of CARD’s impact on overall vaccination uptake.

Following the pandemic, the public health unit involved in one of the cluster trials⁹ sought to re-introduce CARD into the school vaccination program to try to improve vaccine acceptance. Institutional memory of CARD, however, was not retained due to large staff turn-over after the pandemic. To improve feasibility of implementation, CARD was integrated using a stepwise approach that commenced with jurisdictional schools with lower vaccination uptake rates. This approach provided an opportunity to examine the specific effects of CARD on vaccine acceptance as uptake of three targeted vaccines could be compared over time and between schools where CARD was implemented and those where it was not.

The present manuscript reports on the results of this real-world study. The study hypothesis was that CARD would increase vaccine uptake rate. If demonstrated to be effective, CARD could be used as the framework for vaccination delivery to not only improve vaccination experiences, but also to improve vaccine uptake, and in turn, reduce vaccine-preventable disease.

Materials and methods

Design, setting and Participants

We undertook a partially blinded, controlled before and after (pre-post) study involving schools with grade 7 students (12 years old) receiving vaccination services from Wellington-Dufferin-Guelph Public Health (WDG) in Ontario, Canada. Grade 7 is the age targeted for school-based vaccinations in Ontario. Public and Catholic schools were ranked by the public health unit according to vaccination uptake observed in the 2022–2023 grade 7 cohort of students using a median split. The bottom 29 schools were allocated to CARD (intervention group) and the remaining 38 to control (standard care/usual practice) for the subsequent year’s grade 7 cohort of student’s vaccinations (2023–2024). This resulted in an approximately equal number of students in each group.

All public health staff who work in the school vaccination program in WDG have regular education around vaccines provided by the organization as part of their role. CARD training was added and delivered by the program manager and content experts. CARD training involved a half-day workshop with all public health staff involved in school vaccinations in July 2023, to educate them about CARD and to review how specific interventions (e.g., student education ahead of time, environmental interventions for allocated vaccination spaces) would be integrated into work processes for CARD schools, and data collection procedures. All public health staff were therefore aware of the intervention. Staff provided vaccination services to both intervention (CARD) and control schools. Students eligible for vaccination and parents/caregivers (herein, parents) were unaware of the study and hypothesis.

Vaccination consent and process

In both groups, the public health unit sent an e-mail to the schools with an e-mail template to send to parents which included a link to an online vaccine consent form and resources (fact sheets and videos) on all three vaccines routinely offered by public health at school [i.e., Human papilloma virus (HPV), Hepatitis B (HB), and Meningococcal conjugate-ACYW vaccine (MCV)].¹² In CARD schools, this e-mail also included information about CARD and links to CARD resources. The school sent the e-mail to parents. Then robocalls were sent to parents in all schools 2–3 days prior to the clinic as a reminder of vaccination day.

Nurses made two visits to each school to carry out vaccination clinics: one in the fall (2023) and one in the spring (2024). The order of vaccine injection was standardized in both groups, and included MCV first, HB second, and HPV last for fall (2023) clinics. Spring (2024) clinics included the second (and final) dose of the series for HB and HPV, with HPV administered second. Nurses documented vaccination details using usual electronic methods. Additional study procedures are described below according to the study group. Data collection tools were adapted from prior studies, with demonstrated reliability (>90% inter-rater agreement) and construct validity (sensitivity to change, and ability to discern between groups).^{9–11,13–16}

CARD group

CARD interventions mimicked those in our prior school-based vaccinations studies (see also www.cardsystem.ca).^9–11 Briefly, a public health nurse contacted school principals to plan vaccination clinics, including scheduling a visit prior to scheduled fall vaccination clinics to educate and prepare students for upcoming vaccinations. Nurses delivered an in-class multi-media presentation that included content about the vaccines being offered and how to use the CARD acronym (C-A-R-D) to cope before, during, and after vaccination. Children were given a paper copy of a CARD coping checklist and asked to select strategies from the different letter categories that they wanted to use for their upcoming vaccinations. Examples included having a support person present, using a distraction item and being injected in private. They also specified their baseline level of fear of needles and history of fainting on the checklist.

Nurses collected completed CARD checklists to assist with preparation of upcoming vaccination clinics. For instance, on clinic day, the order of vaccination was determined by level of fear of needles (most fearful first) and coping preferences (e.g., vaccination in a private room away from peers, and/or use of a topical anesthetic cream to reduce pain from injection). Nurses also applied environmental changes on clinic day to reduce fear cues (e.g., separating waiting and vaccination areas, arranging furniture so that students faced distractions rather than medical equipment, and obscuring needles). Students were invited to confirm their self-selected coping strategies on vaccination day in fall and spring clinics. The same completed checklist was used to guide interactions during fall and spring clinics; students could modify their responses. Nurses supported students in their preferred coping options during injection(s).

Immediately after vaccine injections, students completed an optional paper-based feedback survey. Questions inquired about whether CARD helped them accept being vaccinated at school, and the helpfulness of CARD. They also reported on helpfulness of the CARD checklist and nurse in-class education specifically. Separately, nurses reported their perceptions of CARD’s helpfulness for student coping and whether CARD altered the satisfaction and duration of their interactions using a paper survey.

Control group (usual care)

Nurses contacted principals to plan upcoming vaccinations; however, they did not routinely visit schools prior to vaccination day to educate and prepare students. On vaccination day, nurses reviewed all vaccines with the students prior to administration. Students were not invited to self-select preferred coping strategies from the CARD checklist. However, as the same nurses provided care to CARD and non-CARD schools, some environmental and process changes may have been transferred from CARD to non-CARD schools. Students and nurses did not complete feedback surveys after vaccination.

Approval for the project was granted by the Research Ethics Board of the University of Toronto. A waiver of consent was granted to allow for population-level data collection.

Sample size and analysis

The primary outcome was change in vaccination uptake between 2023–2024 (after CARD intervention) and 2022–2023 (before) within the CARD group, which was assessed for the three targeted vaccines. For each school calendar year (i.e., 2022–2023 and 2023–2024), the data reflect vaccinations administered between September 1 to August 31. Secondary outcomes included change in vaccination uptake within the control group, differences in uptake between CARD and control groups, and change between years in all schools together (CARD and control). The study sample included all grade 7 students in the 2023–2024 school calendar year eligible for vaccination in jurisdictional schools. Based on the prior year’s data (2022–2023), this was about 3500 students. Using the observed uptake rates, the sample size allowed us to detect an absolute difference of about 3.5% or more.¹⁷

We conducted an intent-to-treat analysis, which included all grade 7 students in included schools, regardless of whether individual students participated in the CARD intervention or not. Data were summarized using descriptive statistics. Vaccine uptake was compared between and within groups using Pearson’s chi squared test. A post-hoc sensitivity analysis was conducted that repeated the analyses after combining 2021–2022 and 2022–2023-years’ data. A p-value of 0.05 was considered statistically significant. Data were analyzed using Stata 15.0 (College Station, Texas).

Results

Altogether, all 67 public and Catholic schools that receive vaccination services from WDG Public Health were included. Twenty-nine were selected for CARD integration and 38 served as controls. The percent of public schools was higher in the CARD group (82.8%) compared to control (57.9%); p = .03. The average number of grade 7 students per school was 48 (SD = 22) in 2022–2023 and 49 (23) in 2023–2024, with no differences between groups: p = .16 and .77, respectively.

Table 1 displays within group percentages and differences in vaccination coverage before (2022–2023) and after (2023–2024) CARD integration in selected CARD schools and control schools. Within the CARD group, 2023–2024 vaccine uptake was significantly higher than 2022–2023 for all vaccines: HPV (+9.8%; p < .001), HB (+8.1%; p < .001), and MCV (+5.3%; p < .001) respectively. Within the control group, vaccine uptake rates decreased or stayed the same: HPV (−1.1%; p = .50), HB (−4.4%; p = .004), and MCV (−2.7%; p = .03).

Table 1.

Within group percentages and differences in vaccine coverage before (2022–2023) and after (2023–2024) CARD integration in CARD and Control schools.

Vaccine	2022–2023 (Before)†	2023–2024 (After)†	Percent Difference	p-value*
CARD schools	(n = 1478)	(n = 1403)
Human papillomavirus (HPV)**	809 (54.7)	905 (64.5)	+9.8%	<.001
Hepatitis B (HB)**	891 (60.3)	959 (68.4)	+8.1%	<.001
Meningococcal conjugate-ACYW (MCV)	1133 (76.7)	1150 (82.0)	+5.3%	<.001
Control schools	(n = 1729)	(n = 1877)
Human papillomavirus (HPV)**	1136 (65.7)	1213 (64.6)	−1.1%	.50
Hepatitis B (HB)**	1242 (71.8)	1265 (67.4)	−4.4%	.004
Meningococcal conjugate-ACYW (MCV)	1485 (85.9)	1562 (83.2)	−2.7%	.03

^†Values are frequency (%); *Chi squared test; **Two doses required for series completion.

Table 2 displays between group percentages and differences in vaccination coverage before (2022–2023) and after (2023–2024) CARD integration in selected CARD schools and control schools. Vaccine uptake was significantly lower (p < .001 for all analyses) for all 3 targeted vaccines in schools selected for CARD integration at baseline. However, after CARD integration, there was no significant difference between groups for any vaccine (i.e., CARD schools caught up to controls).

Table 2.

Between group percentages and differences in vaccine coverage before (2022–2023) and after (2023–2024) CARD integration in CARD and Control schools.

Vaccine	CARD schools†	Control schools†	Percent Difference	p-value*
Before Phase (2022–2023)	(n = 1478)	(n = 1729)
Human papillomavirus (HPV)**	809 (54.7)	1136 (65.7)	−11.0%	<.001
Hepatitis B (HB)**	891 (60.3)	1242 (71.8)	−11.5%	<.001
Meningococcal conjugate-ACYW (MCV)	1133 (76.7)	1485 (85.9)	−9.2%	<.001
After Phase (2023–2024)	(n = 1403)	(n = 1877)
Human papillomavirus (HPV) (HPV)**	905 (64.5)	1213 (64.6)	−0.1%	.94
Hepatitis B (HB)**	959 (68.4)	1265 (67.4)	+1.0%	.56
Meningococcal conjugate-ACYW (MCV)	1150 (82.0)	1562 (83.2)	−1.2%	.35

^†Values are frequency (%); *Chi squared test; **Two doses required for series completion.

When vaccine uptake was compared in all schools together before (2022–2023) and after (2023–2024) CARD integration, a significant increase was observed for HPV uptake in the after phase: 64.6% vs. 60.6%; p = .001. There was no difference for HB (67.8% vs 66.5%; p = .27) and MCV (82.7% vs. 81.6%; p = .27), respectively. Post-hoc sensitivity analysis did not lead to substantive changes in the results.

Table 3 summarizes student and nurse feedback during the fall (2023) and spring (2024) school vaccination clinics in the CARD group. Across both clinics, about one-third of students said that CARD helped them consent to receiving vaccinations at school and two-thirds reported that CARD helped them during vaccinations. More students reported that the CARD checklist helped than the nurse education lessons. Most nurses reported that using CARD helped student coping and that it did not change satisfaction or duration of interactions with students compared to usual practice.

Table 3.

Student and nurse feedback in the CARD group for 2023–2024 vaccinations.

	Fall 2023 clinic*	Spring 2024 clinic*
Students†
No. reporting CARD positively impacted consent to vaccination (%)	311 (36.7)	262 (31.0)
No. reporting CARD helped during vaccination (%)	579 (69.4)	551 (66.5)
No. reporting CARD checklist helped (%)	312 (36.6)	302 (35.3)
No. reporting class education helped (%)	242 (28.4)	121 (14.1)**
Nurses‡
No. reporting CARD helped student	727 (85.4)	727 (86.4)
No. reporting no change in satisfaction of their student interaction	649 (76.3)	705 (85.5)
No. reporting no change in duration of their interaction	575 (67.0)	701 (82.1)

Values are frequency (percent); *Sample ranges from n = 834 to n = 858 respondents.

^†Student survey questions.

Some people say that CARD helps them to say yes to getting vaccinated at school. Did CARD help you to say yes to getting vaccinated at school today? Response options: i) yes, ii) no, iii) I don’t know. Responses were dichotomized (yes vs. all other). Positive (yes) responses are provided.

We use the CARD (Comfort Ask Relax Distract) system to help make needles less scary and painful. Tell us how much the CARD system helped with your vaccination today. Response options: i) not at all, 2) a little bit, 3) a medium amount, 4) a lot. Responses were dichotomized (not at all vs. all other). Responses for all other (i.e., CARD helped) are provided.

Tell us all the things that helped you with your needle today. i) The class presentation from the nurses, 2) Choosing what you wanted to do from the CARD checklist. Responses were closed-ended (yes/no). Positive (yes) responses are provided:

**This question was retained, however, only one class education session was held (prior to fall clinic).

^‡Nurse survey questions.

How much did CARD help? Response categories: i) not at all, ii) a little bit, iii) a medium amount, iv) a lot. Responses dichotomized (not at all vs. all other). Responses for all other (i.e., CARD helped) are provided.

How satisfied were you with your interaction relative to normal? Response options: i) less satisfied, ii) same, iii) more satisfied. Responses for same are provided.

How long did the interaction take relative to normal? Response options: i) less time, ii) same, iii) more time. Responses for same are provided.

Discussion

Schools are an efficient and accepted setting for vaccinating youth.¹⁸ Many students, however, experience pain, fear and associated immunization stress-related responses^1–3 Concerns about these sequelae can contribute to vaccine refusal.⁵ The CARD system is an evidence-based framework for vaccine delivery that incorporates interventions that reduce pain, fear and other immunization-stress related responses, and improves the experiences of individuals undergoing vaccination. In this study, jurisdictional schools within WDG with lower vaccination uptake rates in the 2022–2023 school calendar year were allocated to receive CARD for the 2023–2024 school calendar year; the remainder received usual care. In CARD schools, uptake for all targeted vaccines was significantly higher in 2023–2024 compared to 2022–2023; the absolute increase ranged from 5.3% to 9.8%. In contrast, there was no change or a slight decrease in control schools. CARD schools caught up to controls in coverage of all vaccines in 2023–2024, with no statistically significant differences between groups. Overall, there was an absolute increase in HPV uptake of 4.0% in 2023–2024 compared to 2022–2023. About one-third of students reported CARD positively contributed to their decision to be vaccinated at school and most students and public health nurses reported that CARD helped students during vaccinations. The majority of nurses reported that vaccination interactions resulted in a similar level of satisfaction and lasted the same amount of time when compared to usual practice.

This study expands on prior research demonstrating the benefits of CARD on vaccine recipient vaccination symptoms and experiences, including vaccinations in schools, mass vaccination pop-up clinics, community pharmacies and hospitals.^{9–11,13–16,19–21} Importantly, this is the first study to examine the specific impact on population-level vaccination uptake. The observed benefit of CARD on vaccine uptake may be due to multiple factors, including educating and preparing students ahead of time, building relationships and trust between students and providers, and providing vaccination services using a person-centered approach. Together, these interventions address multiple aspects of vaccine acceptance related to complacency, confidence and convenience.²² The results are also consistent with prior studies whereby vaccine recipients that knew about CARD ahead of time reported that it positively influenced their decision to be vaccinated.^15,16,20

The benefit of CARD on overall HPV vaccine uptake is worthy of further discussion. All vaccines target different vaccine preventable diseases and are not necessarily judged as equal by caregivers and students. That the overall HPV vaccine uptake rate was increased might be because concerns held by students and caregivers for this particular vaccine might have been better addressed by CARD, including fear of pain of the vaccine.²³ It is also possible that there was more room for improvement for HPV vaccine due to the lower baseline vaccination rate.

The mode of delivery for student education also warrants further discussion. In all our studies with CARD involving school-based vaccinations, we incorporated a face-to-face lesson delivered by a public health nurse at school.^9–11 For some public health units (not including WDG) a class lesson on the specific vaccines is already incorporated in their school vaccination program and content can be easily adjusted to incorporate CARD. In the current study, WDG added the lesson to their workflow. This required an additional visit to the school and introduced additional costs to the vaccination program that were absorbed by WDG. Public health units such as WDG allocate resources to the school vaccination program according to their needs. WDG elected to implement CARD to improve vaccine uptake in schools with lower uptake rates. This was considered important to promote equity in health across the community they serve. A decision was made to integrate CARD using a stepwise approach to improve feasibility. The effectiveness of other modes of education could be explored in the future, including online resources such as the CARD web game²⁴ and face-to-face lessons delivered by provider trainees such as nursing or pharmacy students. Our prior work also indicates that classroom teachers are willing to educate students about coping with CARD; however, they are not willing to educate about the vaccinations themselves.²⁵ Nurses are required to administer vaccinations and therefore, applying CARD on vaccination day.

As with any practice innovation, organizational resources are also required to train and support staff to incorporate CARD. In this study, we utilized a workshop approach, which included managers and senior nurses to inform adaptations to the local context. We have recently developed an online CARD course to train providers.²⁶ The e-module includes information about the rationale and impact of CARD and key practice tools to support implementation. Its development was informed from our learnings from this and other projects of the need for a single comprehensive training resource. We recommend the e-module as the first step to CARD implementation to streamline training and improve organizational readiness for CARD.

This study has some limitations. First, schools were assigned to study groups according to baseline vaccination uptake (i.e., lower and higher) and some of the observed change may be due to regression to the mean (i.e., extreme values/outliers likely to normalize). In re-introducing CARD after the pandemic, the focus was on trying to raise vaccine uptake rates across low vaccine uptake district schools. Randomizing schools was not deemed appropriate for this goal. Second, the statistical analysis approach assumed independence of observations; however, data were clustered by school. As individual characteristics are not routinely collected, this prevented alternative statistical approaches from being used. Third, this study evaluated vaccination uptake trends over only a short time frame. The public health unit did not track vaccination coverage by school until after 2021, preventing analyses in prior years. Importantly, there was triangulation of the data. Students corroborated the positive impact of CARD on their decision to accept vaccinations, and together with the statistical findings, is highly suggestive of a real effect of CARD. Fourth, one public health unit participated, limiting generalizability of the results. The included district, however, is similar to many others across Ontario.

There are some strengths of the study. First, the design utilized a real-world, population-based approach, which increases external validity. The public health unit prioritized schools with lower vaccination coverage rates for the CARD intervention, which maximized efficiency for their program. This is because there was more room to improve in schools with lower baseline coverage rates. Second, the study included a control group, which allowed examination of the vaccination coverage without the intervention. Third, the analysis utilized an intent-to-treat approach that maintained group allocation regardless of treatment received. This likely contributed to a more conservative estimate of the effect of CARD because some aspects of CARD were not performed with complete fidelity. For instance, face-to-face education was missed by students that were absent from school on the day of the lesson. Fourth, vaccination status was recorded by nurses in real-time in an electronic database, which is considered an accurate way of tracking vaccinations. Finally, students and parents were unaware of the study hypothesis, minimizing threats to validity due to measurement bias.

A prior review of school-based vaccinations urges evidence-based processes for optimizing school-based vaccination acceptance.² Multi-component and multi-level interventions have been specifically recommended to bolster vaccinations, including educational interventions, reminders, and vaccine mandates.²⁷ It is important to note, however, that the observed effects of educational interventions that have been employed across studies, as well as their impact on vaccine acceptance, vary widely, with some studies demonstrating no effect.^28–31

The findings of this study have important implications for public health units delivering vaccinations at school. In all instances, public health seeks interventions that improve the uptake of vaccines being offered to students. HPV is one vaccine that has consistently lagged behind other vaccines offered by WDG in terms of acceptance. The present study added the systems-level CARD educational intervention to ongoing vaccination reminders and a long-standing vaccine mandate specifically for MCV.^30,32 The results showed that CARD can improve vaccine uptake, including for HPV. Given the promising results of this study, other public health units can adopt CARD as the framework for vaccination delivery to promote more positive vaccination experiences for students as well as to try to improve vaccination uptake and reduce the risk of vaccine-preventable diseases.

While the results were positive, however, there continues to be room for improvement in coverage for all targeted vaccines. Additional studies are recommended to confirm the results, including in diverse geographical regions and populations. Additional research can examine the effectiveness of CARD according to other factors that affect vaccine uptake, as well as the connection to uptake of COVID-19 vaccines. Research is also recommended to examine whether benefits accrue over time. As students, parents and school staff become familiar with CARD and share stories about positive experiences with CARD, it may lead to higher vaccine acceptance.

In summary, vaccine uptake was increased in schools with lower uptake that use CARD in one public health unit in Ontario. Together with other research, the results of the present study further emphasize the value of CARD as a framework for vaccination delivery in school-based vaccinations.

Biography

Anna Taddio is professor, Leslie Dan Faculty of Pharmacy, University of Toronto, Senior Associate Scientist, The Hospital for Sick Children, and GSK Research Chair in Vaccine Education and Practice-Oriented Tools. Her program of research examines the short-term and long-term effects of pain and pain management, the effectiveness and safety of pain management interventions, and evidence-based practice and implementation research.

Funding Statement

This study was funded by a Public Health Agency of Canada Immunization Partnership Fund award (2324-HQ-000142) and a Canadian Institutes of Health Research Foundation Grant (FRN 159905) awarded to A. Taddio. The funding agencies had no input into the study.

Disclosure statement

A. Taddio reports a University of Toronto Section 9 Trademark No. 924835 for CARD™. A. Taddio is inaugural Glaxo-Smith-Kline (GSK) Research Chair in Vaccine Education and Practice-Oriented Tools.

LMB reports grants from the World Health Organization (WHO), Public Health Association of British Columbia (PHABC), and Merck Canada unrelated to this work.

CMM reports a Genome Canada grant unrelated to this work.

Author contributions

AT, KM, LMB were involved in conception and design. AT, KM, VG, LMB, CL, CMM, LTW, NEM were involved in analysis and interpretation of the data. AT wrote the first draft of the paper. AT, KM, VG, LMB, CL, CMM, LTZ, NEM were involved in revising it critically for intellectual content. All authors provided final approval of the version to be published, and all authors agree to be accountable for all aspects of the work.