Study design

A cluster randomised controlled trial with measurements at baseline and at follow-up after 3 months. Cluster randomisation was used since the randomisation was first conducted at school level and thereafter classes were randomised to be included. Follow-up after 12 and 24 month are not presented in this paper.

Setting

Sweden is a multicultural country: almost a third of all children under the age of 18 years have an immigrant background.28 The Swedish upper secondary school, which the vast majority of students attend, conducts both theoretical and vocational programmes that reach out to adolescents aged 16–19 years. According to Swedish law, all students should have access to school health. The school health work is mainly preventive and involves at least a school nurse and a school physician; however, only the nurse works in the school on a regular basis. In Sweden, sexual education is mandatory in primary and secondary school, and includes topics such as anatomy, sexuality, prevention of sexually transmitted infections and reproductive health.

All first year upper secondary school students (aged 16 years) are offered a health interview with the school nurse, who provides a dialogue regarding psychosocial health, eating habits, sleep, physical activity, intake of tobacco, alcohol and drugs, as well as sexual health and relationships. This intervention is optional, although usually all students do participate in it. The interview is scheduled for approximately 1 h and conducted in an empathic atmosphere based on Motivational Interviewing29 that focuses on the individual student's health and well-being.

Population and sample

First year upper secondary school students attending the regular health interview with the school nurse in the autumn semester of 2014 were eligible for participation. We excluded students who were not able to speak or write Swedish (ie, recently arrived immigrants), and adolescents with severe learning disabilities and development disorders (ie, studying at special schools). Upper secondary schools (n=18) in nine municipalities in central Sweden a total of about 600 000 rural and urban inhabitants who represented different socioeconomic levels-were included. The participating schools included both municipally and privately managed schools that offered vocational as well as theoretically-oriented education, and had a varying number of students.

Recruitment and randomisation

School nurses were recruited to the project via the school heads and through direct contact at a national school health conference. For logistical reasons, those working in the far north or south of Sweden were not invited. Initially 59 upper secondary schools were approached and eventually 18 of these were included; details of the recruitment are given in the flow chart in figure 1. A total of 23 school nurses working in 20 schools in 9 municipalities agreed to participate in the study. Three nurses in two schools dropped out at the start of the intervention due to heavy workload. This opportunistic selection resulted in a total of 20 school nurses working in 18 schools.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Flow of schools and students through trial.

Randomisation was performed in two steps. First, in order to avoid contamination, the schools were randomised into either the intervention group or the control group. The schools were randomly drawn by administrative personnel not involved in the project. Second, 113 school classes within these schools were randomly selected to be included in the study; this number was chosen in order to achieve the desired number of students according to the power calculations (described below). The school nurses provided basic details about the classes (eg, Social science 14A), and number of students in each class. If the class consisted of less than 25 students, an additional class was allocated. The students were recruited by the school nurses, as described below. The end result was an intervention group of 394 students from 60 classes and a control group of 357 students from 53 classes (figure 1). Recruitment did not begin until registration was made public.

Two school nurses in two schools dropped out during the intervention due to termination of employment and personal reasons, and they did not complete the health interviews with their allotted students. In order to compensate for these losses, which were all from the intervention group, all school nurses in the intervention group were advised to perform health interviews with an additional class. Consequently, similar classes were included in the study.

The students (n=832) were invited to participate in the study when they met the school nurse for the general health interview. Those who agreed to participate (n=751) gave informed written consent. Before the health interview started, all students were asked to complete a baseline questionnaire. A follow-up questionnaire was completed after 3 months (n=741). The baseline questionnaire was completed individually at the school nurse's office while the follow-up questionnaire was given to the whole class. Students not present at this time could complete the follow-up questionnaire afterwards at the school nurse's office. School nurses were provided with checklists regarding the procedure and used log lists (protocols completed after each health interview) to assure that the intervention was performed in a uniform fashion. For obvious reasons, the school nurse could not be blinded to whether the student belonged to the intervention or control group, but the research assistant who recorded the data from the participants did not possess this knowledge.

Theoretical framework

The Health Belief Model (HBM) was used as a theoretical framework. This model has previously been used in studies about HPV and HPV vaccination,30 ,31 and in interventions with the aim to increase prevention of sexually transmitted infections.19 ,32 According to the HBM framework, a person's health behaviour can be explained by the individual's beliefs regarding health actions. HBM includes the following central constructs: perceived susceptibility, perceived severity, perceived benefit and perceived barriers. Furthermore, sociodemographic factors, such as age, sex, ethnicity and parental education level, as well as knowledge, are recognised as factors that can indirectly influence the individual's behaviour. The main limitation of the model is that it does not consider emotional or relational aspects involved in decisions regarding health behaviour.33

Intervention

The intervention was included in the general 1 h face-to-face health interview. Thus, while the controls received general information, including those on sexual health, the intervention group received specific HPV education guided by HBM and a predesigned informational structure format. The school nurse showed a specially designed flipchart with pictures and brief information to the students (see example in table 1). She also handed out a specially designed leaflet. The intervention took about 30 min and included the following information:

• General facts about the virus;

• Viral transmission;

• What HPV can cause;

• Risk factors;

• Prevention, that is, safe sex with condom use and HPV vaccination;

• Locations where the girls could receive the vaccine free of charge in the municipality;

• Facts about the HPV vaccine;

• The importance for girls to attend future cervical cancer screening controls.

The leaflet consisted of 12 pages that included similar information and also an HPV quiz, as well as links to the national online youth clinic, the homepage of the university where the researchers worked and contact information to the authors. After the follow-up questionnaire was completed, students in the intervention group were provided with condoms. Students in the control group only received standard treatment, that is, the regular health interview, as described above.

Outcome measures

The outcomes pertain to the individual student. Primary outcomes: intention to use condom with a new partner and beliefs towards primary prevention about HPV (strongly agree to strongly disagree, see online supplementary appendix).

Secondary outcomes: increased HPV vaccination (yes/no/do not know) and increased condom use (yes/no).

Pilot study

In early 2014, the intervention procedure, including the educational material and questionnaire, was tested by three school nurses among 45 students aged 16 years. The nurses' experiences resulted in minor revisions: for logistic reasons the intervention procedure was modified to having the follow-up for all students together in the classroom, instead of having individual follow-ups at the school nurses office. Furthermore, the educational material was shortened, some statements were simplified and the additional response alternative ‘do not know’ was added to the block of questions regarding beliefs about HPV. A few of the questions were clarified; two were considered redundant and therefore these were removed. In parallel, the questionnaire was tested among adolescents aged 17 years (n=230)34 and they confirmed the school nurses’ observations.

Instrument

The questionnaire was based on previous research and clinical experience. The questions about beliefs, awareness and knowledge about HPV (n=24) were adapted from our previous study,15 and these had multiple choice alternatives and six-point verbal rating scales (Likert scale) from ‘Totally agree’ to ‘Totally disagree’, including ‘Do not know’. The demographic background questions (n=14) were taken from the national questionnaire for adolescents,35 and the questions regarding sexual behaviour (n=17) were based on a project carried out on university students that has been used repeatedly since 1989.10 The questions about beliefs towards primary prevention of HPV according to the HBM constructs on susceptibility comprised questions regarding the risk of contracting HPV; while severity included questions on how serious it would be to receive an HPV infection or cancer. In addition, questions about benefits comprised confidence in vaccine effectiveness and intention to vaccinate, while barriers embraced the individual's perceived barriers for HPV vaccination such as fear of needles and difficulties encountered while booking an appointment for vaccination.

Validity and reliability of the intervention, including the questionnaire

The validity and reliability were rigorously tested with both qualitative and quantitative methods. Two focus group interviews were undertaken with adolescents (n=8) aged 16–17 years (both boys and girls) who were asked what they considered important to include in an intervention regarding prevention of HPV. Cognitive interviews (n=5) and discussion sessions (n=8) regarding the questionnaire were performed with adolescents aged 15–18 years.

To test the stability and reliability of the intervention over time, a test–retest evaluation was undertaken in 2014 with first year upper secondary school students (n=29) randomly selected from a school in a city of 200 000 inhabitants that was located in mid-Sweden. The questionnaire was distributed on two occasions with a time interval of 2 weeks. Analysis was based on Cronbach's α and showed high reliability scores for the questions regarding the decision-making process relating to HPV vaccination (0.800–0.998), except for the single question ‘Do you want to be vaccinated later’ (0.436). Statements regarding beliefs about HPV ranged from low to high (scores 0.331–0.918); consequently, two statements were removed.

Ethical considerations

The study was conducted according to the Declaration of Helsinki. All participants received oral and written information before giving their written consent. The participants were informed that participation was voluntary, that they could withdraw participation at any time without providing a motivation or incurring any negative consequences for themselves. They were also informed that only the researchers would have access to the data, and that all data would be presented at a group level. Contact details to the researchers were provided in case of further questions. According to the Swedish law, children above 15 years of age who understand what participation means have the right to give informed consent regarding participation in research studies.36 Therefore, informed consent was not obtained from the parents. We asked permission to conduct the study from the head of the school health facility in each municipality and from the principals of the schools.

Education to school nurses

All participating school nurses (n=20) received written and verbal instructions, and participated in educational sessions (with MG and CS) scheduled for about 2 h. The education comprised factual information about HPV and the HPV vaccine. The flipchart educational material was presented, and the nurses were encouraged to give comments if anything was unclear or if anything they considered as important was missing. Furthermore, each school nurse received a minimum of 1 h additional education at the time for the start of the intervention at the school where she worked. During the intervention, the nurses were contacted on a weekly basis.

Sample size calculation

The power calculation was based on a previous study of the research group37 and clinical experience. The sample size of 400 participants per study arm were based on assumptions of a baseline intention to use condom if new partner of 60%, with a power of 80% to detect differences of 10 percentage points between intervention and control group at a significance level of 5% (356/study arm IG/CG, a dropout of 10% and missing values=400).

Statistical analysis

For descriptive statistics, categorical data are presented as frequencies and percentages, n (%); ordinal data as medians, means and SDs; while continuous data are given as mean and SD. Differences between the intervention and control groups are tested with Pearson’s χ² test for categorical data, Mann-Whitney test for ordinal data, and Student's independent samples t test for continuous data.

The effects of the intervention were measured from baseline at follow-up. First, the HBM scores were calculated by measuring the difference between baseline and at follow-up for each HBM question. Then the questions were grouped together according to the HBM constructs susceptibility, severity, benefits and barriers. Finally, the total HBM index was calculated by adding all HBM constructs together. For each individual item in the HBM index, the answers on the 5-point Likert scale (‘strongly agree’ to ‘strongly disagree’) where scored 0 to 4 for negative questions (ie, when ‘strongly disagree’ implied higher health beliefs while ‘strongly agree’ implied lower health beliefs) and 4 to 0 for positive questions (ie, when ‘strongly disagree’ implied lower health beliefs while ‘strongly agree’ implied higher health beliefs). ‘Do not know’ was classified as an answer for the neutral option. Finally, scores from all included individuals were summarised to give a total score. The McNemar test was used to determine differences in actual HPV vaccinations and actual condom use from baseline to follow-up; we have excluded the ‘do not know’ group for vaccination in the analysis. The analyses are conducted at an individual level and are based on intention-to-treat principle; these comprise all the 741 students completing the follow-up questionnaire.

In order to take into account the dependence between students who were informed by the same school nurse, generalised estimating equations models were used for examining the results of the intervention on the outcome measures. The differences between the outcome variables at baseline and follow-up were used as dependent variables in the generalised estimating equations models, while the treatment group (intervention or control) was used as a predictor together with the socioeconomic and demographic variables, sex and immigrant background, which differed significantly between the two groups at baseline. These variables were only included as main effects in the regression models, with no interaction being used. In all the analyses, a two-sided p value <0.05 was considered as statistically significant. All analyses were performed in IBM SPSS Statistics V.22.0.