We examined whether the effect of an intervention recommending HPV vaccination for mothers with daughters had persisted 4 months after the intervention, and whether the degree of persistence of the intervention effect depended on the type of message: a statistical message only, a patient’s narrative in addition to statistical messages, or a mother’s narrative in addition to statistical messages.
The results of the present study showed that vaccination intention in the all groups slightly increased, although those changes were not significant. The effect of intervention had not persisted at 4 months after intervention, and the degree of persistence of the intervention effect 4 months after intervention did not differ among the three message types. More precisely, vaccination intention 4 months after intervention decreased to a level that did not significantly differ from the level before intervention under all three intervention conditions. Thus, the answer to our research question 1 was “no”. However, in the group that received a patient’s narrative plus statistical messages, the difference of vaccination intention between before and 4 months after intervention was marginally significant. The number of participants in the present study was smaller than that in our previous study because of loss to follow-up. We conducted post-hoc power analyses; the statistical power in the patient’s narrative plus statistical messages condition was 55%. The analysis may have been underpowered to detect the modest difference that would be expected if there were an appropriate sample size.
The amount of change in vaccination intention 4 months after intervention did not differ significantly among the three message types. Thus, the answer to our research question 2 was also “no”. Our previous study showed that vaccination intention increased significantly directly after reading intervention materials, and that statistical messages plus a mother’s narrative increased vaccination intention the most . However, the present findings revealed that those effects wore off during the 4 months following intervention. Our results are consistent with those of a previous study showing that risk perception, intention to be vaccinated, and HPV vaccine uptake did not differ significantly between baseline and 3 months following intervention among female university students in the United States .
HPV vaccination expects multiple injections given over a series of weeks. In Japan, the second dose is given 1 month after the first dose, and the third dose is given 6 months after the first dose. Studies indicate that there are a number of anti-HPV vaccination websites on the internet  and that seeing anti-vaccination messages online can negatively affect the audience’s attitude toward vaccination, even 5 months after exposure to these messages . Considering this, persistently increased vaccination intention after an intervention to promote HPV vaccines is important, to effectively conduct vaccination.
To maintain the intervention effect, intervention messages should be easy to remember so that they can be recalled over time by the message recipients . Studies indicate that narratives are more easily remembered than didactic content, and narratives are the preferred mental structure for storing and retrieving information [15,16]. Events and characters in narrative messages are linked to each other through personal, causal, temporal, and spatial associations, which facilitate storage and retrieval of more complex information because the recipient need only remember a single story rather than miscellaneous information . Studies indicate that health materials with narrative messages enhance recall and are associated with lower decision conflict than messages without narratives [18-20]. However, the advantage of narratives for memory and recall did not seem to be present in this study, considering that the intervention effect of narratives in addition to statistical messages diminished over time to the same degree as the effect of statistical messages only. The reason may be that the narratives used in the present study were created for research purposes and thus may have been too short and uninteresting to be remembered and recalled over time by participants. It may be useful to examine the persuasiveness of a longer and more vivid narrative of an experience of cervical cancer in future studies, as the narrative of a patient or a mother is more likely to be retained and recalled and will consequently influence decision making regarding vaccination over time.
A review of studies about media exposure indicate that the degree of exposure to health messaging among message recipients is positively associated with the likelihood of their engagement in health behaviors . It is considered that repeated exposure to health messages enhances recognition and recall of messages among recipients and encourages them to undertake certain health behaviors . To improve persistence of the intervention effect of messages recommending HPV vaccination, it may be important for health institutes to frequently disseminate messages such that target populations are repeatedly exposed to those messages. However, the mean vaccination intention directly after intervention was less than 3 (i.e., somewhat unlikely) in our previous study . Frequent messaging alone may be insufficient to increase vaccine uptake. Messaging coupled with other interventions, such as facilitating access to the vaccine and providing educational programs, may help to improve vaccine uptake.
Additional studies indicate that educational programs on topics such as anti-stigma toward mental illness and self-management of chronic illness have improved participant's knowledge, attitude, and practices over time [23-25]. Although no studies have examined the effect of educational programs in the context of HPV vaccination as far as we know, provision of educational programs that communicate the benefits and risks of HPV vaccines in a balanced manner may contribute to persistently increased vaccination intention among participants. Especially in Japan where negative attitudes toward HPV vaccines focusing on the risk of vaccination have been established, educational programs that provide a set of facts and narratives and clarify misunderstandings regarding HPV vaccines may be more influential for participants’ long-term decision making than only disseminating written messages.
Several limitation should be considered in this study. First, when the persuasive intent is obvious, narrative persuasion is hindered because some audiences may resist if they feel they are being manipulated . This constraint may be related to the negative results of the present study and should be noted, in addition to the brief content of the materials, as discussed above. Second, we did not examine the duration of the intervention effect, i.e., how many weeks did the intervention effect persist? Investigating this issue is necessary so as to determine the appropriate frequency of message exposure to sustain increased vaccination intention. Third, we assessed vaccination intention rather than vaccine uptake. However, behavioral intention is generally measured in public health studies because it predicts an actual behavior . Fourth, whether some participants had their daughter(s) receive HPV vaccines after intervention is unknown. However, the influence of this on the study results is considered to be small because the HPV vaccination rate is only a few percent in Japan. Fifth, factors other than intervention, such as exposure to media coverage, may have influenced vaccination intention among some participants during the 4 months after intervention. Sixth, the response rate in the present study was 57%; the intention of the 43% of participants who did not respond to the follow-up survey are unknown. The respondents in the present study could be self-selected in the follow-up survey. Selection bias may have influenced the study results. Finally, the sample size in this follow-up study was smaller than that in our initial study. The analyses in the present study may have been underpowered to detect the differences that would be expected if there were appropriate sample sizes.