Emergency contraceptive pills in Sweden: evaluation of an information campaign

Authors


Dr M. Larsson, Department of Women's and Children's Health, Uppsala University, S-751 85 Uppsala, Sweden.

Abstract

Objective  To evaluate a community-based intervention regarding emergency contraceptive pills, including a mass media campaign and information to women visiting family planning clinics.

Design  Quasi-experimental.

Setting  Two counties in Sweden.

Population  Eight hundred randomly selected women aged 16–30 years, 400 women in the intervention group and 400 in a comparison group.

Methods  Postal questionnaires before (2002) and after (2003) the intervention.

Main outcome measures  Exposure to the intervention, knowledge, attitudes, practices and intention to use emergency contraceptive pills.

Results  Before the intervention, the response rate was 71% (n= 564); after the intervention, the corresponding figure was 83% (n= 467); overall response rate 58%. Two-thirds (64%) of the targeted women had noticed the information campaign. One out of six who had visited a family planning clinic during the intervention year recalled being given information about emergency contraceptive pills. Specific knowledge and attitudes improved over time in both groups, but there was no difference in change between the groups. The proportion of women who had used emergency contraceptive pills increased from 27% to 31% over time. Intention to use emergency contraceptive pills in case of need was reported by 74% of the women and remained stable over time, but logistic regression showed that information during the previous year contributed to willingness to use the method in the intervention group.

Conclusions  Knowledge, attitudes and practices about emergency contraceptive pills increased in both groups. Emergency contraceptive pills is gradually becoming a more widely known, accepted and used contraceptive method in Sweden, a trend that may have limited the impact of the intervention.

INTRODUCTION

Prescription free progestin-only emergency contraception was introduced in Sweden in 2001, and sales figures increased by 34% from 2001 to 2002. In 2002, 91% of the total volume of doses (n= 94,469 out of 103,812) was distributed over-the-counter without prescription.1 Youth clinics and family planning clinics continue to distribute the product for free on request because some women may find it difficult to afford the cost of 13 Euros.

It has been argued that a more widespread use of emergency contraception could affect the abortion rates.2–6 This has not yet been demonstrated in Sweden, where approximately every fourth pregnancy leads to an induced abortion and the abortion rate according to official statistics was 18.7 per 1000 women in 2001, 19.6 per 1000 women in 2002 and 20.3 per 1000 women during the first six months of 2003.7,8 In order to use emergency contraceptive pills properly, women need to be aware of the existence of this option and know enough about it not to have unfounded worries. They need to be able to obtain the product preferably within 24 hours and they also need sufficient funds for the purchase. Previous Swedish studies have shown that most women have good awareness of the existence of emergency contraceptive pills, and in both quantitative and qualitative studies predominantly positive attitudes toward emergency contraceptive pills have been demonstrated.9–12 The most recent study showed, however, that in spite of general awareness and positive attitudes, one-third of the women considered emergency contraceptive pills to be an abortion. It was also demonstrated that correct knowledge and positive attitudes relating to emergency contraceptive pills contributed to willingness to use the method in case of need.13

According to a recommended research agenda proposed by Glasier in 2002,14 one area of interest is the effectiveness of emergency contraception as a public health strategy to reduce unwanted pregnancies. Improving access, raising knowledge and providing advance supply of emergency contraceptive pills are possible components in this strategy. Several trials directed towards selected groups have been reported to be successful,15–17 but only a few community-based interventions regarding emergency contraception have been published so far. Trussell et al.18 demonstrated that a media campaign increased both the awareness of the method and the number of calls to a Hotline. In Mexico, a three-year programme of training for health care providers and a multifaceted media campaign increased awareness of emergency contraception among both providers and clients.5 In order to reach all parts in a community, it has been argued that programmes should employ a combination of strategies. The most effective strategy may be to involve one-on-one interventions for high risk individuals, community-wide interventions attempting to change social norms, and finally, policy-level efforts to help modify the social and political environment.19 Programmes intended to influence health behaviour should also preferably be based on theory.20

On the basis of these recommendations and our previous findings that misunderstandings and doubtful attitudes about emergency contraceptive pills still exist in spite of the deregulation, we decided to undertake a community-based intervention aimed to increase knowledge of, and improve attitudes towards, emergency contraceptive pills so that intention to use the method in case of need could increase.

METHOD

The intervention was founded on the Theory of Diffusion of Innovation and the Health Belief Model. The Diffusion Theory was developed in 1983, and is defined as the process by which an innovation (in this case emergency contraception) is communicated through certain channels over time among the members of a social system.21 The process typically includes five stages: innovation development, dissemination, adoption, implementation and maintenance. The Health Belief Model developed in the 1950s has been widely used as a guiding framework for health interventions.21 The main concept applicable to our intervention is the likelihood of taking action (intention to use emergency contraceptive pills) in case of need, which was one of the main outcome measures in our study. The intervention took place during one year from 1 April 2002 to 31 March 2003, and was designed to combine different strategies.

The campaign consisted of the following parts:

  • A media campaign with the main message that emergency contraceptive pills is a good emergency solution and a specially designed brochure containing basic facts about emergency contraceptive pills such as: content, mechanism of action, recommended time frames for use, side effects and efficacy if taken on the first, second or third day after unprotected intercourse. This information was spread via different channels; advertisements in the local newspaper and the student newspaper, posters in schools, universities and student clubs and information on the rear of local buses during two periods of two weeks and one of three weeks. The brochure was delivered to the waiting room of all school-nurses, gynaecologists, youth clinics and family planning clinics in the intervention area.
  • Nurse–midwives working in family planning clinics were informed about the intervention and were asked to provide oral and written information about emergency contraceptive pills during the intervention period. This request was passed on to the nurse–midwives on several occasions and each clinic received written information about the study. The brochure was supposed to be delivered to women coming for contraceptive counselling or for follow up after childbirth.
  • Information on a website including recent research findings concerning emergency contraceptive pills.
  • Women requesting induced abortion were offered one package of emergency contraceptive pills to keep at home.

We mailed questionnaires together with a cover letter to evaluate the effect of the intervention on two groups of women aged 16 to 30 shortly before and after the intervention. Participants were randomly selected from the national tax registration record and stratified for postcode area. The sample consisted of 400 women in one county (the intervention group) and 400 women in another county (the comparison group). The two counties are situated in mid-Sweden, both containing one medium-sized city with about 130,000 inhabitants and surrounding rural areas. The sample size was calculated to detect a 10% difference between the intervention group and the comparison group at a significance level of P < 0.05 (two-tailed test) regarding knowledge of time frames for use of emergency contraceptive pills. This knowledge was shown to be lacking in 60% of women in a previous study.9 We also assumed a response rate at follow up around 50%. The questionnaires were coded in order to identify non-respondents who received three reminders at three-week intervals. One year later, the women who had answered the pretest received a similar questionnaire. Two reminders were sent.

The pre-campaign questionnaire consisted of 24 questions that have been presented in an earlier study.13 Five multiple-choice questions concerned demography. One question explored whether the women had received any contraceptive counselling during the previous year. Six multiple-choice questions with two to eight response alternatives concerned knowledge of emergency contraceptive pills, awareness of emergency contraceptive pills, source of information, knowledge of mechanism of action, effectiveness if taken on the first day or third day after intercourse and side effects. Seven verbal descriptive rating scale questions explored attitudes regarding emergency contraceptive pills and over-the-counter availability, potential side effects, impact on regular contraceptive use, need for individual counselling, and whether or not emergency contraceptive pills could be considered to be an abortion. The rating scale questions had five response alternatives ranging from ‘totally agree’ to ‘totally disagree’. Three multiple-choice questions with two to five response alternatives investigated the women's previous use and estimated future use of emergency contraceptive pills. One dichotomous question explored if they favoured a pharmacy or a clinic for future purchase of emergency contraceptive pills. Two open-ended questions about the motives for their choice and their general opinion of emergency contraceptive pills ended the questionnaire.

At follow up, we used 22 of the same multiple-choice questions and added another two multiple-choice questions with four and six response alternatives in order to investigate whether or not the women had noticed the campaign about emergency contraceptive pills during the previous year. The local Medical Ethics Committee at Uppsala University gave their approval of the study.

Data were entered and analysed on the Statistical Package for Social Sciences, SPSS 11.0. The mean age of respondents was compared with that of non-respondents using the t test. The answers to the questions regarding knowledge were categorised in two categories: correct answer = 1 and incorrect answer/don't know = 0. The answers to the four questions regarding knowledge were added to form an index of knowledge, which could add up to a maximum of four points if all answers were correct. Differences between the intervention group and the comparison group were tested at baseline and at follow up with Fisher's exact test and Pearson χ2 test for two independent samples on nominal scaled variables and with the Mann–Whitney method for ordinal scaled variables. Differences within the groups from baseline to follow up were tested using the McNemar test for two related samples on nominal scaled variables and the Wilcoxon Mann–Whitney U test for ordinal scaled variables. Differences in change between the intervention group and the comparison group were tested with the χ2 test. Differences were considered significant if P < 0.05.

To examine possible predictive factors for the dependant variable ‘estimated future use’ (yes/no, don't know), a logistic regression model was performed. All the explanatory variables, except age, were dichotomised before inclusion in the model.

RESULTS

The response rate for the pretest was 70.5% (n= 564) and for the posttest 82.8% (n= 467). The overall response rate for women who participated in both tests was 58.4%, higher in the intervention group (65.3%) than in the comparison group (51.5%). An analysis of the dropouts showed that they did not differ from the respondents with respect to age, mean age for respondents and for non-respondents (23.7 and 23.0, respectively; P= 0.097), but differed in ethnic origin and present occupation (Table 1). The completion rate of the questions was high, ranging from 98.5% to 100%.

Table 1.  Analysis of respondents (R) versus non-respondents (NR).
 R (n= 467)/NR (n= 97), %P
  • 1

    Fisher's exact test.

  • 2

    χ2 test.

Non-Nordic origin7/200.0011
Living with partner45/410.5751
Living with children12/120.8641
Student51/50 
Employed41/330.0122
Unemployed2/8 
Comprehensive school17/23 
Upper secondary school54/520.3682
University29/26 
Intervention group56/470.1451
Comparison group44/53 

The mean age of the women was 23.7 years (range 16–30) and the intervention group did not differ from the comparison group (P= 0.156). The main demographics are shown in Table 2 and show that the intervention group and the comparison group were similar with respect to ethnic origin, educational level and cohabiting status. The groups differed both at baseline and at follow up regarding living with children (P= 0.000/0.002) and regarding present occupation at follow up (P= 0.005).

Table 2.  Main demographics of the intervention group and the comparison group before the intervention.
 Intervention group (n= 261), %Comparison group (n= 206), %
Non-Nordic origin77
Living with partner4348
Living with children718
Student5546
Employed3844
Unemployed23
Comprehensive school1520
Upper secondary school5553
University3128

A majority of women in the intervention group recalled that they had received some kind of information during the intervention period as opposed to the women in the comparison group where only a minority remembered any kind of information activities (Table 3). Newspapers and magazines were most often cited in both groups. All the information channels used in the media campaign were more often noted by women in the intervention group, whereas the information channel not used in the campaign, TV, was equally often noted by the women in the comparison group. The most cited ‘other’ information channel was the Internet.

Table 3.  Information about the emergency contraceptive pill during the previous year in the intervention group and the comparison group.
 Intervention group (n= 261), %Comparison group (n= 206), %P
  • 1

    Fisher's exact test.

Received information about emergency contraceptive pills6429<0.0011
Newspapers and magazines4421<0.001
Buses180<0.001
TV8100.506
Posters162<0.001
Other1020.001

Almost half of the women had visited a family planning clinic during the previous year but the majority stated that they had not received any information about emergency contraceptive pills on this occasion. There was no difference between the intervention group compared with the comparison group (Table 4).

Table 4.  Information about emergency contraceptive pills from clinics during the previous year in the intervention group and the comparison group.
 Intervention group (n= 261), %Comparison group (n= 206), %
Visited clinic during previous year4647
No information8487
Oral information88
Written information30
Oral and written information55

The awareness of emergency contraceptive pills was high, almost all women, 97.1–99.6% in both groups, had heard of emergency contraceptive pills. The intervention group had better knowledge of the method already at baseline. This better knowledge remained after the intervention. There was no difference in change between the groups except for knowledge about side effects, which had decreased over time in the comparison group (Table 5). There was an overall improvement over time measured by the knowledge index (P= 0.028). The improvement from 2.03 to 2.17 in the intervention group reached significance (P= 0.028), but not the improvement from 1.69 to 1.78 in the comparison group (P= 0.467). A comparison of the difference in change over time did not show any significant difference (P= 0.370). The attitudes toward the method remained mostly stable, but for three of the statements about emergency contraceptive pills there was a change towards more favourable attitudes over time in the intervention group. A comparison of the change over time between the two groups showed, however, no difference (Table 6).

Table 5.  Knowledge of emergency contraceptive pills; percentage of correct answers in the intervention group and the comparison group.
 Intervention group (n= 261, before/after), %Comparison group (n= 206, before/after), %Difference of intervention group vs comparison group at baseline (P)Change over time (intervention group/comparison group) (P)Difference in change over time (P)
  • 1

    Fisher's exact test.

  • 2

    McNemar's test.

  • 3

    χ2 test.

Mechanism of action61/6557/590.34310.222/0.61420.4363
Effectiveness day 166/7150/590.00110.093/0.02020.7883
Effectiveness day 325/2817/200.03910.342/0.42920.1463
Side effects52/5347/410.30510.919/0.14220.0383
Table 6.  Attitudes to emergency contraceptive pills in the intervention group and the comparison group; percentage (%) of women who agreed with the statements.
 Intervention group (n= 261, before/after), %Comparison group (n= 206, before/after), %Difference of intervention group vs comparison group at baseline (P)Change over time (intervention group/comparison group) (P)Difference in change over time (P)
  • 1

    Mann–Whitney's U test performed on the five-point rating scale.

  • 2

    Wilcoxon's signed-rank test performed on the five-point rating scale.

  • 3

    χ2 test.

Prescription free emergency contraceptive pills is good79/8576/780.31910.030/0.21320.9193
Emergency contraceptive pills is a kind of abortion30/2637/310.00810.026/0.05420.4423
Hesitation to use because of side effects25/2023/200.20810.041/0.26520.7813

The use of emergency contraceptive pills had increased in both groups over time but we found no difference between the groups. The willingness to use emergency contraceptive pills in the future remained stable, as shown in Table 7. To further evaluate the factors associated with willingness to use emergency contraceptive pills in the future, we performed a multiple logistic regression model and included all demographic characteristics, the group assignment, whether or not the women had noticed the information campaign, previous use of emergency contraceptive pills and their willingness to use emergency contraceptive pills at baseline as possible explanatory variables. Only the opinion held at baseline and previous use of emergency contraceptive pills were found to be influencing factors. But when we split the participants in the two groups and included the same possible explanatory variables except group assignment, information during the previous year was also found to be one of the influencing factors for the willingness to use emergency contraceptive pills in the future in the intervention group (Table 8).

Table 7.  Use of emergency contraceptive pills in the intervention group and the comparison group.
 Intervention group (n= 261, before/after), %Comparison group (n= 206, before/after), %Difference of intervention group vs comparison group at baseline (P)Change over time (intervention group/comparison group) (P)Difference in change over time (P)
  • 1

    Fisher's exact test.

  • 2

    McNemar's test.

  • 3

    χ2 test.

Ever used emergency contraceptive pills29/3324/290.17410.022/0.00620.3611
Would use emergency contraceptive pills in case of need76/7674/720.36310.775/0.40220.8353
Table 8.  Multiple logistic regression model of factors influencing future use of emergency contraceptive pills in the intervention group and the comparison group. Values are expressed as odds ratio (95% CI).
VariablesIntervention groupPComparison groupP
  • *

    Dichotomized non-Nordic (reference) versus Nordic.

  • **

    Dichotomized living with children (reference) versus not living with children.

  • ***

    Dichotomized living with partner (reference) versus single.

  • Dichotomized other occupation (reference) versus student.

  • ††

    Dichotomized 12 years or more of education (reference) versus less than 12 years.

Age0.8260.471
Ethnicity*1.42 (0.39–5.19)0.5933.90 (0.88–17.30)0.074
Living with children**1.13 (0.35–3.60)0.8411.45 (0.54–3.93)0.462
Cohabiting status***1.93 (0.87–4.25)0.1041.37 (0.54–3.50)0.507
Occupation0.86 (0.37–1.98)0.7151.05 (0.41–2.70)0.919
Education level††3.54 (0.72–17.41)0.1201.05 (0.27–4.17)0.942
Positive to future use at baseline10.18 (4.66–22.23)<0.00111.39 (4.99–25.96)<0.001
Ever used emergency contraceptive pills3.50 (1.29–9.49)0.0146.57 (1.62–26.62)0.008
Information about emergency contraceptive pills previous year2.39 (1.14–4.99)0.0210.85 (0.36–2.00)0.709

DISCUSSION

We believe this study to be the first attempt to evaluate a mass media campaign about emergency contraception using not only a pre- and post-intervention design, but also including a comparison group. The study showed some improvement over time in both the intervention group and the comparison group, indicating a secular trend in the whole community towards better knowledge, more favourable attitudes and an increased use of emergency contraceptive pills. This trend may have erased the impact of the intervention, because although the media campaign about emergency contraceptive pills was observed by two-thirds of women targeted, it did not have the power to increase the knowledge or improve the attitudes to emergency contraceptive pills in the intervention group beyond what was found in the comparison group. The part of the intervention focussing on routine information to individuals by nurse–midwives was not very successful. Only 16% of the 120 women in the intervention group who had visited a family planning clinic during the previous year recalled being given any information about emergency contraceptive pills. The logistic regression model showed, however, that information during the previous year was one of the influencing factors for willingness to use emergency contraceptive pills in the intervention group.

One of the main challenges with this quasi-experimental design is the possibility of selection bias. The sample was stratified for age and postcode area and the two groups were similar with respect to all the demographic variables at baseline, except that women in the comparison group had children to a higher extent. This is not likely to have influenced the result according to the logistic regression model. The overall response rate of 58% is a matter of concern, but the dropout rate was equally distributed in both groups. We do not have much information about the 29% of the women who did not return the baseline questionnaire. We know that their age did not differ from the age of the respondents but the percentage of immigrant women was lower (11.7%) than in the population as a whole in these two counties (15% and 17%). We have more information about the women who dropped out at follow up (Table 1), and although many similarities existed between respondents and non-respondents, differences were found with regard to ethnicity and occupation. This difference, however, was found in both the intervention group and the comparison group, so we do not think it had any importance for the result of the evaluation.

Many sexual health interventions have shown limited or no effect. This has been discussed in a number of review articles19,20,22,23 and in the following section we will examine some of the possible reasons for limited impact which were put forward by Merze and d'Afflitti.19

  • Methodological issues: We have already discussed some of the sampling issues. We used a longitudinal cohort sample, which may have the shortcoming of suffering from a high dropout rate with the most motivated individuals likely to continue participation. Another reason for the lack of effect may be low statistical power. Due to financial and logistic limitations, our study included only one pair of communities. Some evaluators recommend at least 10 communities in order to be able to detect differences.
  • The influence of secular trends: If a general shift occurs in society, it may erase the effect of an intervention (i.e. significant changes take place in both intervention and comparison communities over time). This has been demonstrated in a number of smoking prevention trials19 and is possibly an explanation of our findings of changes over time in both groups in our study. One alternative explanation of this might be that the study itself alerted the participating women to learn and reflect more about the topic of investigation. However, we believe that the improvement of the knowledge index in the intervention group is the result of a combination of a secular trend and our intervention.
  • Magnitude of effect: The results of a number of community health programmes indicate that an effect of an intervention should be expected to be relatively small; many changes were less than 5%.19 But it has also been argued that even small changes on a community level can have an important public health impact, as so many individuals are involved. It was impressive to find that almost every woman in our study was already aware of the existence of emergency contraceptive pills, a level of general knowledge that had reached the ‘roof’ and thus could not be improved by the intervention. The more specific knowledge summarised in the knowledge index had increased somewhat over time. This holds promise for the future and shows that even without expensive information campaigns, emergency contraceptive pills will slowly become a more accepted and more widely known contraceptive method.
  • Limitations of the intervention: It was surprising and disappointing that our efforts to engage nurse–midwives in routine information about emergency contraceptive pills failed. A previous study had shown that 72% of nurse–midwives agreed that routine information about emergency contraception should be included in contraceptive counselling and 41% reported that they often or always informed women coming for contraceptive counselling about emergency contraceptive pills.24 One possible explanation of the lack of effect of this part of the intervention may be the length of the intervention. Changing routines takes time and one year may have been too short a period to implement this additional strategy. Another explanation might be insufficient tailoring. Nurse–midwives were not given any extra time or resources, except for the information brochure to distribute. Counselling sessions are often restricted to 15–30 minutes and additional routines might therefore have been impossible to include. We did not provide any research protocols for the nurse–midwives to fill in because we wanted their situation to be as close to everyday life as possible. Finally, it is of course possible that information about emergency contraceptive pills was given to the consulting women, but because it was not requested by the women, they quickly forgot about it. This highlights an important issue for clinicians. To what extent is it meaningful to engage in health promotion activities in connection with consultations? Prochaska and Di Clemente have developed a transtheoretical model for behaviour change which is described as a gradual, continuous, dynamic process in which people move through a sequence of five stages: precontemplation, contemplation, preparation, action and maintenance.21 They claim that counselling must begin with the assessment of the actual stage of every individual in order to be successful. Routine information, given to everybody without this reflection, may therefore not be cost effective.
  • Limitations of theory: Our intervention was influenced by theories about health behaviour and community development such as the Health Belief Model and the Diffusion of Innovation Theory. According to the Health Belief Model, the likelihood of action is a result of perceived benefits minus perceived barriers. We assumed that fear of side effects and the opinion that emergency contraceptive pills is a kind of abortion could be possible barriers for intention to use emergency contraceptive pills. It is therefore promising that these attitudes had changed over time in the intervention group. Our results indicate that emergency contraceptive pills in Sweden has reached the third stage in the Diffusion theory; adoption, since practically all women were aware of the method, but it is not yet fully implemented in routine contraceptive counselling and in women's contraceptive practices. Our study was not designed to test the theories and all concepts and constructs were not taken into account. It is possible that a more rigorous use of the theories would have influenced the effect of the intervention. On the other hand, this was not the aim of the study and the inclusion of other theoretical concepts would have resulted in a more extensive questionnaire, probably resulting in a lower response rate.

When we compare our findings with the more promising results from other media campaigns,5,18 it is important to note that inclusion of a comparison group strengthens the validity of the findings. We could easily have concluded that the intervention was more successful if we had not added a comparison group, as there was a significant change over time in the intervention group on a number of outcome measures. Our intervention did not have the ‘advantage’ of starting from a very low level of knowledge as in the American campaign.18 On the contrary, awareness about emergency contraceptive pills was shown to be very high already at baseline and attitudes to, and intention to use emergency contraceptive pills were also fairly good. We did not have the opportunity to have ‘earned’ (free) media attention as was reported by the Mexican example.25 Emergency contraception is not a controversial method in Sweden any longer and therefore does not attract much attention in different media.

In spite of the increased use of emergency contraceptive pills in Sweden, by nearly one-third of women in this study, there is no sign of impact on abortion rates.26 One obvious question is then, of course, “why not?” We have repeatedly joined in the chorus arguing that better knowledge, more positive attitudes and an increased use of emergency contraception have the power to reduce the number of induced abortions. This has not yet been proved, and we have to admit that emergency contraception does not seem to be an obvious solution. The battle against unwanted pregnancies must continuously be fought on many different arenas—sexual education in schools, empowering of women, youth-friendly and culturally sensitive counselling services and development of, and easy access to, effective and well-tolerated contraceptive methods for both women and men.

CONCLUSION

Emergency contraceptive pills is gradually becoming a more widely known, accepted and used contraceptive method in Sweden and both groups had improved their knowledge and attitudes about emergency contraceptive pills and used it more after the intervention. This secular trend in the whole community may have limited the impact of the intervention. Engaging health personnel in information campaigns is a demanding task that requires careful planning and allocation of resources. Widespread knowledge about, and easy access to, emergency contraception is one of many components in the prevention of unintended pregnancies.

Acknowledgements

The authors would like to thank Patrik Öberg at the Swedish University of Agricultural Sciences and Sylvia Olofsson at the Uppsala Clinical Research Center for statistical advice. The study was funded by grants from Uppsala County Council, The Family Planning Fund of Uppsala and the Swedish National Institute of Public Health.

Accepted 3 April 2004

Ancillary