Impact of visit characteristics on intention to recommend HPV vaccine: An experiment with US health care professionals

Abstract

Objective.

Presumptive recommendations that assume parents want to vaccinate can increase human papillomavirus (HPV) vaccine uptake. We sought to examine how visit characteristics affect health care professional’s (HCP) intention to use this evidence-based recommendation style.

Methods.

In 2022, we conducted an online experiment with 2,527 HCPs who had a role in adolescent vaccination in the United States. Participants read 1 of 8 randomly assigned vignettes about a well-child visit. Using a 2 x 2 x 2 between-subjects factorial design, the vignettes varied the following visit characteristics: patient age (9 vs. 12-year-old), prior parental vaccine refusal (yes vs. no), and time pressure on the HCP (low vs. high). HCPs reported on their intention to use a presumptive HPV vaccine recommendation, as well as on related attitudes, subjective norms, and self-efficacy. Analyses used 3-way analysis of variance and parallel mediation.

Results.

Participants were pediatricians (26%), family/general medicine physicians (22%), advanced practitioners (24%), and nursing staff (28%). Overall, about two-thirds of HCPs (64%) intended to use a presumptive recommendation. Intentions were higher for older children (b=0.23) and parents without prior vaccine refusal (b=0.39, both p<0.001). Time pressure had no main effect or interactions. HCPs’ attitudes and self-efficacy partially mediated effects of patient age and prior vaccine refusal (range of b=0.04-0.28, all p<0.05).

Conclusion.

To better support visits with younger children and parents who have refused vaccines, HCPs may need more training for making presumptive recommendations for HPV vaccine. Reinforcing positive attitudes and self-efficacy can help HCPs adopt this evidence-based recommendation style.

INTRODUCTION

Presumptive recommendations are a key strategy for increasing uptake of human papillomavirus (HPV) vaccine among children. Presumptive recommendations are characterized by direct statements that assume parents want to vaccinate. For example, a health care professional (HCP) can say, “Now that your child is 9 years old, he’s due for a vaccine to protect against HPV cancers. We’ll give it at the end of the visit.” Multiple randomized clinical trials have found that training HCPs to use presumptive recommendations like this one increases HPV vaccine uptake,^1,2 while observational studies further indicate that such recommendations are associated with lower levels of HPV vaccine refusal and delay.³ Based on this research, using presumptive recommendations is recognized as an evidence-based strategy for improving persistently lackluster uptake of HPV vaccine in the United States (US), where only half (50%) of 13-year-old children complete the multi-dose vaccine series,⁴ despite national guidelines for routine vaccination by age 12.⁵ Unfortunately, despite the effectiveness of presumptive recommendations, less than two-thirds (42%-65%) of HCPs report consistently using them for HPV vaccine.^3,6 Findings from observational studies vary even more widely,^7-9 with one study estimating the use of presumptive recommendations to be as low as 15% for HPV vaccine.⁹ These data suggest that additional work is needed to support HCPs in adopting presumptive recommendations and sustaining their use across the wide range of patient encounters they may have with age-eligible children.

Understanding when HCPs choose presumptive recommendations could help inform future interventions aimed at improving consistent use. Although few existing studies have examined the circumstances of presumptive recommendation specifically, research in HPV vaccine communication more broadly suggests HCPs vary their communication by clinical visit characteristics in several ways. First, HCPs recommend the vaccine more often for older versus younger children,^10,11 which appears to be due, in part, to the routine immunization schedule’s focus on HPV vaccination at ages 11-12.⁵ This focus may discourage vaccination at the earliest possible ages of 9-10, even though it is allowed and actively encouraged by professional organizations including the American Academy of Pediatrics and the American Cancer Society.^12,13 Second, HCPs express greater willingness to make timely HPV vaccine recommendations to parents who they believe to be supportive of vaccination.¹⁴ Finally, HCPs indicate that the short length of clinical visits can be a barrier to HPV vaccine communication,^15-17 which could make direct strategies like presumptive recommendations especially valuable when they are running behind schedule. In these ways, research suggests that older child age, no prior parental vaccine refusal, and high time pressure are circumstances in which HCPs may be more likely to use presumptive recommendations.

Research is also needed to understand why clinical visit characteristics impact HCPs’ use of presumptive recommendations. Previous studies have applied the Theory of Planned Behavior¹⁸ to understand HPV vaccine recommendation practices,^19,20 suggesting that HCPs’ perceptions, including attitudes, subjective norms, and perceived behavioral control, guide their intentions to engage in communication behavior, such as using presumptive recommendations. For example, HCPs may be more inclined to use presumptive recommendations with parents of older children because they perceive those parents as more accepting (an attitude), their colleagues as more approving (a subjective norm), and themselves as more capable of delivering the recommendation (perceived behavioral control). Identifying mechanisms, like these perceptions, that support presumptive recommendations in various contexts could further inform future interventions to improve the consistency of their use.

To address these research gaps, we conducted an experiment to examine which visit characteristics influence presumptive recommendations and the mechanisms by which these characteristics do so. We hypothesized that HCPs’ intention to use presumptive HPV vaccine recommendations is greater in clinic visits where they expect a positive environment in terms of parents’ receptivity to vaccines or potential time savings (Hypothesis 1). Thus, we predict higher use in the context of older patient age, no prior parental vaccine refusal, and higher time pressure. Furthermore, we hypothesized that these characteristics interact such that the positive effect of patient age on intention is stronger with no prior parental vaccine refusal and under high time pressure than with vaccine refusal and under low time pressure (Hypothesis 2). Finally, drawing on the Theory of Planned Behavior, we hypothesized that the predicted effects of the experimental manipulations (older patient age, no prior parental vaccine refusal, and higher time pressure) lead to more positive Theory of Planned Behavior constructs (attitudes, subjective norms, and self-efficacy), which are in turn associated with higher intention to use presumptive recommendations (Hypothesis 3). By understanding when and why HCPs choose presumptive recommendations, our study aims to inform future interventions seeking to promote HCPs’ consistent use of this evidence-based communication style when recommending HPV vaccination.

METHODS

Participants

Participants were physicians, advanced practitioners (advance practice nurses and physician assistants), and nursing staff (certified nursing assistants, licensed practical/vocational nurses, medical assistants, and registered nurses). Eligible HCPs worked in pediatric, family medicine, or general practice in the US and had a role in HPV vaccination for children ages 9-12. We estimated a minimum total sample size of 2,500 participants to obtain 80% power to detect a 10% effect size over the current 50% prevalence, given 2-tailed tests and a critical alpha of 0.05.

We sampled HCPs from an online panel managed by WebMD Market Research, which verifies licenses of physician and advanced practitioner members. To construct the study panel, WebMD assessed interest in study participation among members with eligible clinical roles (physicians, advanced practitioners, and nursing staff) via email. The survey company then emailed a study invitation to 6,278 members with up to 4 reminders. We applied quotas to sample similar proportions of pediatricians, family physicians, advanced practitioners, and nursing staff, and we oversampled participants in rural counties (as defined by US Department of Agriculture Rural-Urban Continuum Codes 4-9),²¹ given the rural-urban disparities in HPV vaccination coverage.^22,23 A total of 2,527 HCPs completed screening questions to verify their eligibility, provided informed consent, and completed the survey, which yielded a 57% response rate (American Association for Public Opinion Research formula 3).²⁴

Procedure and measures

We conducted an online experiment with a 2 x 2 x 2 between-subjects factorial design in May-July 2022. Participants read a vignette that described a well-child visit with a child named Michael and his mother. We randomly assigned participants to 1 of 8 versions of the vignette that experimentally varied 3 visit characteristics: time pressure (on time with vs. 45 minutes behind on patient schedule), patient age (9-year-old vs. 12-year-old), and prior parental vaccine refusal (always accepted vaccines vs. sometimes refused vaccines, Table 1). Vignettes ended with a description of a presumptive recommendation: “One way to recommend HPV vaccine is to start with a direct approach that assumes parents want to vaccinate. You might say: ‘Today, Michael is due for a vaccine to protect against HPV cancers. We’ll give it at the end of the visit.’”

Table 1.

Experimental manipulation of visit characteristics

Visit characteristic	Vignette component
Time pressure	Imagine that it’s a typical day in your clinic. You’re currently [on time with/45 minutes behind on] your patient schedule.
Patient age	Your next patient is Michael, a [9-year-old/12-year-old] boy, who is attending a well-child visit with his mother.
Prior parental vaccine refusal	Michael has been your patient since he was a toddler, and his mom has [always accepted vaccines/sometimes refused vaccines] you have recommended for him.
	Today, Michael is eligible to get his first dose of HPV vaccine.

After reading the assigned vignette, participants indicated how much they agreed with 4 statements developed based on the Theory of Planned Behavior.¹⁸ One item measured intention to use presumptive recommendation in the vignette presented (“I would use a direct approach like this one to recommend HPV vaccination to parents like Michael’s mom”). The other 3 items measured attitudes about whether using a presumptive recommendation would lead to vaccine uptake (“Parents like Michael’s mom usually accept HPV vaccine if recommended with a direct approach”), subjective norms about whether colleagues would approve using presumptive recommendation (“My closest colleagues approve of using a direct approach to recommend HPV vaccine to parents like Michael’s mom”), and perceived behavioral control in terms of self-efficacy to use a presumptive recommendation (“I feel confident that I could use a direct approach to recommend HPV vaccine to parents like Michael’s mom”). All items had a 5-point response scale (“Strongly disagree” [coded as 1] to “Strongly agree” [5]).

Our survey also assessed participants’ demographic and clinical characteristics. Demographic characteristics were clinical role, years in practice, gender, and race/ethnicity. Clinical characteristics were practice type, healthcare system membership, rurality (by county and state of participants’ practices), and US Census region. We conducted cognitive interviews with 8 HCPs, who were not survey participants, to assess and improve comprehension of the vignette and survey items prior to fielding.

Participants who provided written informed consent and completed the survey received an electronic gift card worth $30-$45, depending on their clinical role. The Institutional Review Board at the University of North Carolina at Chapel Hill approved this experiment. We registered this experiment with AsPredicted (#96686) before data collection.

Data analysis

We used a 2 x 2 x 2 between-subjects factorial analysis of variance (ANOVA) to examine the effects of the 3 experimental manipulations of visit characteristics (patient age, prior vaccine refusal, and time pressure) on HCPs’ intention to use a presumptive recommendation (Hypotheses 1 and 2). We present the effect size of each main and interaction effect as partial eta-squared (η²), which describes the proportion of total variance explained by a given variable after accounting for other variables in the model. A sensitivity analysis probed the experimental effects by rerunning the same ANOVA separately for participants who prescribe vaccines (i.e., physicians and advanced practitioners) and nursing staff (many of whom can administer vaccines when their clinics have standing orders). Because an examination of Levene’s test suggested unequal variance for some main effects, we reexamined each main effect with a mixed linear model and found similar data patterns.

To examine why the visit characteristics had their effects, we used mediation analyses (Hypothesis 3), applying MacKinnon’s approach²⁵ that is similar to a causal mediation model.²⁶ This approach involved computing specific indirect effects via 3 parallel mediation models. We entered the experimentally manipulated visit characteristics (patient age, prior vaccine refusal, and time pressure) and mediating variables (attitudes, subjective norms, and self-efficacy) into each model. We assessed the statistical significance of indirect effects using the bootstrap method with 5,000 repetitions. We report unstandardized path coefficients (b) and standard errors (SE). Statistical tests were 2-tailed with a critical alpha of 0.05. We conducted ANOVAs using PROC GLM and PROC MIXED functions in SAS version 9.4 (Cary, NC) and mediation analysis with model 4 of PROCESS version 4.3 (Hayes, 2023).

RESULTS

Our sample had similar proportions of pediatricians (26%), family/general physicians (22%), advanced practitioners (24%), and nursing staff (28%, Table 2). Most participants reported their race/ethnicity as white (66%), Asian (14%), Black (5%), or Hispanic (4%). A majority identified as women (72%). Most participants’ clinics were solo or group practices (61%), part of a healthcare system (62%), and located in non-rural settings (91%). Across clinical roles, the proportions of white and female HCPs in our sample were similar to the population²⁷ (Supplemental Table 1).

Table 2.

Distribution of HCP participant characteristics, United States, 2022 (n = 2,527)

	n (%)
Demographic characteristics
Clinical role
Pediatrician	666 (26)
Family/general physician	557 (22)
Advanced practitionera	603 (24)
Nursing staffb	701 (28)
Years in practice
0-9	950 (38)
10-19	740 (29)
20 or more	837 (33)
Gender
Woman	1,810 (72)
Man	637 (25)
Otherc	80 (3)
Race/ethnicity
White	1,664 (66)
Asian	356 (14)
Black	123 (5)
Hispanic	100 (4)
Multiple races/ethnicities	94 (4)
Otherd	190 (8)
Clinical characteristics
Practice type
Solo or group practice	1,534 (61)
Othere	993 (39)
Part of healthcare system
Yes	1,564 (62)
No	963 (38)
Rurality
Non-rural	2,295 (91)
Rural	232 (9)
US Census region
South	841 (33)
Midwest	576 (23)
Northeast	505 (20)
West	605 (24)

^aIncludes advance practice nurses (n=405) and physician assistants (n=198).

^bIncludes certified nursing assistants (n=11), licensed practical/vocational nurses (n=64), medical assistants (n=84), and registered nurses (n=542).

^cIncludes nonbinary or another gender (n=10) and “prefer not to say” (n=70).

^dIncludes American Indian or Alaska Native (n=10), Middle Eastern or North African (n=19), Native Hawaiian or Pacific Islander (n=4), other race/ethnicity (n=13) and “prefer not to say” (n=144).

^eIncludes hospital- or academic-affiliated practice (n=512), Federally Qualified Health Center (n=272), state or local health department (n=37), local, community, or non-profit organization (n=116), and “other” (n=56).

Impact of visit characteristics on intention

Overall, two-thirds of HCPs (64%) reported intention to use a presumptive HPV vaccine recommendation (29% strongly agree, 35% somewhat agree), while about one-third (36%) did not intend to use it (8% strongly disagree, 19% somewhat disagree, 9% neither). The 12-year-old patient elicited higher intention to use a presumptive recommendation than did the 9-year-old (M = 3.70 [SE = 0.04] vs. M = 3.47 [SE = 0.04], Table 3). This difference was statistically significant (F(1,2519) = 21.23, p < 0.001). Similarly, parents with no history of prior vaccine refusal elicited higher intention to use a presumptive recommendation than those who had sometimes refused vaccines (M = 3.78 [SE = 0.04] vs. M = 3.39 [SE = 0.04], F(1,2519) = 58.20, p < 0.001). Time pressure did not affect intention to use a presumptive recommendation (p = 0.77). The interactions among patient age, prior vaccine refusal, and time pressure were not statistically significant (range of p-values = 0.10-0.80).

Table 3.

Impact of visit characteristics on HCPs’ intention to use a presumptive HPV vaccine recommendation, United States, 2022 (n = 2,527)

Visit characteristic	Hypothesis	M	SE	df	F	η2	p	Levene’s test (p)
Patient age	1			1	21.23	0.01	<0.001	0.05
9-year-old		3.47	0.04
12-year-old		3.70	0.04
Prior parental vaccine refusal	1			1	58.20	0.02	<0.001	<0.001
Yes		3.39	0.04
No		3.78	0.04
Time pressure	1			1	0.09	0.00	0.77	0.70
Low (on time with patient schedule)		3.58	0.04
High (45 minutes behind)		3.59	0.04
Age x Refusal	2			1	1.11	<0.001	0.29
Age x Time pressure	2			1	1.72	<0.001	0.19
Refusal x Time pressure	2			1	0.07	0.00	0.80
Age x refusal x Time pressure	2			1	2.66	0.001	0.10

Note. df = degrees of freedom, M = mean, SE = standard error.

The sensitivity analysis found that physicians and advanced practitioners exhibited the same pattern of findings as the overall sample (Supplemental Table 2). In this subpopulation, two-thirds of participants (66%) reported intention to use presumptive recommendations. Older patient age and no prior vaccine refusal elicited higher intention to use a presumptive recommendation (both p < 0.001), while time pressure had no effect (p = 0.09). In contrast, nursing staff participants exhibited a different pattern of findings. In this group, a similar proportion of participants (60%) reported intention to use presumptive recommendations. Low time pressure elicited higher intention to use a presumptive recommendation than high time pressure (M = 3.56 [SE = 0.07] vs. M = 3.34 [SE = 0.07], F(1,693) = 4.75, p = 0.03), while patient age and prior vaccine refusal had no effect (range of p-values = 0.07-0.59). As in the primary analysis, we found no statistically significant interactions among experimental manipulations in either subpopulation (range of p-values = 0.05-1.00).

Mediation

Patient age.

The main effect of patient age on intention to use a presumptive recommendation (b = 0.23, SE = 0.05, p < 0.001) was mediated by attitudes about whether using a presumptive recommendation would lead to vaccine uptake (b = 0.04, SE = 0.01), and the effect was also mediated by self-efficacy in using a presumptive recommendation (b = 0.08, SE = 0.03, both p = 0.002, Table 4). After accounting for these mediators, the direct effect of patient age remained statistically significant (b = 0.10, SE = 0.03, p = 0.002), consistent with partial mediation. The mediation by subjective norms about colleagues’ approval of using presumptive recommendation was not statistically significant (p = 0.05). Supplemental Figure 1 shows each component of the indirect effects presented in Table 4.

Table 4.

Indirect effects of visit characteristics on HCPs’ intention to use a presumptive HPV vaccine recommendation, United States, 2022 (n = 2,527)

	Mediator
	Attitudes			Subjective norms			Self-efficacy
Visit characteristic	b	SE	p	b	SE	p	b	SE	p
Patient age	0.04	0.01	0.002	0.01	0.01	0.05	0.08	0.03	0.002
No prior parental vaccine refusal	0.21	0.02	<0.001	0.05	0.01	<0.001	0.28	0.03	<0.001
Time pressure	0.009	0.01	0.40	<0.001	0.01	0.89	0.02	0.03	0.38

Note. b = unstandardized regression coefficient for the pathway from visit characteristic through the mediator to the outcome, SE = standard error.

Prior vaccine refusal.

The main effect of prior vaccine refusal on intention (b = 0.39, SE = 0.05) was mediated by attitudes (b = 0.21, SE = 0.02), subjective norms (b = 0.05, SE = 0.01), and self-efficacy (b = 0.28, SE = 0.03, all p < 0.001, Table 4, Supplemental Figure 1). After accounting for the mediators, the direct effect of prior vaccine refusal remained statistically significant (b = −0.15, SE = 0.04, p < 0.001), consistent with partial mediation.

Time pressure.

We found no indirect effects of time pressure on intention to use a presumptive recommendation (range of p-values = 0.38-0.89).

DISCUSSION

HCPs in our national sample reported moderate intentions to use presumptive recommendations in the context of a well-child visit. However, intentions were lower for visits with younger patients and parents who had refused vaccines in the past. These findings, which are consistent with Hypothesis 1, extend prior research that some HCPs avoid a timely discussion of HPV vaccination with parents they fear will be unreceptive.¹⁴ For these parents, some HCPs may choose to recommend more tentatively in a way that presents HPV vaccination as an elective vaccine, rather than part of routine care.^3,6-8 Although some HCPs believe that such open-ended communication will encourage vaccine uptake by building trust,¹⁴ research consistently finds that the opposite is true.^3,6-8,28 Instead, because addressing vaccine concerns boosts parents’ confidence in HPV vaccination,²⁹ HCPs can use communication approaches that pair presumptive recommendations with strategies to effectively ease parental concerns.³⁰ For younger children, HCPs’ concern about parents’ receptivity is possibly further complicated by HCPs’ lack of awareness that the routine immunization schedule supports HPV vaccination at ages 9-10, since ages 11-12 have been the traditional target.⁵ These findings suggest that interventions, such as communication training and provider prompts to vaccinate at age 9,^30,31 may be needed to make presumptive recommendations a routine clinical practice.

Our findings suggest that increasing HCPs’ positive perceptions about presumptive recommendations can increase their intention to use the recommendation style. Most notably, we found that positive attitudes about parents’ receptivity to presumptive recommendations and self-efficacy to deliver such recommendations partially mediated the relationships between two visit characteristics (age and prior vaccine refusal) and intentions. These findings, which are consistent with the Theory of Planned Behavior¹⁸ and Hypothesis 3, can guide future interventions. Improving attitudes could involve correcting HCPs’ tendency to underestimate the extent to which parents value HPV vaccination³² and emphasizing that many parents are willing to accept vaccination after receiving a presumptive recommendation.^1,2 To increase HCPs’ self-efficacy, interventions can include role-play activities and other mastery experiences, as well as social modeling in the form of demonstrations.^1,33 Equally important is the need to develop HCPs’ ability to address parents’ vaccine questions and concerns in those cases in which parents need more counseling than a presumptive recommendation alone. Future interventions could effectively encourage HCPs’ adoption of presumptive recommendations via such approaches that enhance their attitudes and self-efficacy in using this recommendation style.

Interestingly, our experiment did not find that time pressure affected HCPs’ intention to use presumptive recommendations in the overall sample. This finding is contrary to our hypotheses and studies suggesting that short clinical visits influence HPV vaccine communication.^15-17 This absence of an effect of time pressure could reflect HCPs’ motivation to routinely recommend HPV vaccination regardless of clinical visit length, similar to a previous study on childhood vaccine communication.³⁴ In probing the effect of time pressure on different HCP subpopulations, sensitivity analyses suggested that lower time pressure led to higher nursing staff intentions to use presumptive recommendations, which is counterintuitive since prior research shows presumptive recommendations save time.¹⁹ This inverse relationship may be an indication of nursing staff’s lower capacity to recommend vaccination during their workflow, regardless of patient characteristics. Therefore, training on presumptive recommendations should include nursing staff and data showing that presumptive recommendations save time.¹⁹ Despite our nonsignificant finding on time pressure, studies suggest that other characteristics of the clinical environment likely remain influential in HPV vaccine recommendations and uptake,^35,36 and future studies should consider other determinants, such as team workflow, that could affect presumptive recommendation delivery.

Some limitations of our experiment should be noted. First, our experimental design offers insights on when and why HCPs intend to use presumptive recommendations in an experimental setting, but it did not evaluate actual recommendation practices. Future studies are needed to assess changes in perceptions, intentions, and communication behavior over time and in randomized intervention trials set in clinics. Second, our theory-driven mediation analysis focused on a discrete set of mediators of the relationship between visit characteristics and intention. Other individual, interpersonal, and environmental factors not examined in this study may affect HCP behavior in their real-world clinical practice. Third, the generalizability of our vignettes featuring a well-child visit to other contexts such as acute care visits merits further study. Fourth, even with a description of presumptive recommendations that was refined using cognitive interviews, not all participants may be familiar with the recommendation style. Hence our findings could potentially be less representative of how HCPs may actually use presumptive recommendations. Lastly, we conducted this experiment in 2022 when the Covid-19 pandemic periodically continued to challenge routine vaccination. Future studies could consider examining the relationship between visit characteristics and HCPs’ intention to recommend HPV vaccination at later time points.

CONCLUSION

Our experiment indicates a greater need to support HCPs in delivering presumptive recommendations for younger children and parents who have refused vaccines. Increasing HCPs’ adoption of presumptive recommendations should include strategies like reinforcing positive attitudes and confidence in using this evidence-based recommendation style. Strategies such as improving HCPs’ perceptions about parents’ receptivity to presumptive recommendations and conducting role-play activities are promising intervention approaches to enhance HCPs’ attitudes and self-efficacy.

Highlights.

• Presumptive recommendations are an evidence-based way to introduce HPV vaccination

• Our experiment assessed how 2,527 healthcare professionals use such recommendations

• HCPs’ intentions to use presumptive recommendations were lower for younger children

• Intentions were also lower if parents had refused vaccines in the past

• HCPs’ attitudes and self-efficacy partially mediated these associations

Abbreviations:

ANOVA

Analysis of variance

HCP

Health care professional

HPV

Human papillomavirus

SE

Standard error

US

United States