Effect of educational leaflets and questions on knowledge of contraception in women taking the combined contraceptive pill: randomised controlled trialBMJ 1998; 316 doi: https://doi.org/10.1136/bmj.316.7149.1948 (Published 27 June 1998) Cite this as: BMJ 1998;316:1948
- Paul Little (), Wellcome training fellowa,
- Simon Griffin, Wellcome training fellowa,
- Joanne Kelly, research assistanta,
- Nigel Dickson, clinical teachera,
- Carolyn Sadler, family planning trainerb
- a Primary Medical Care, Faculty of Health Medicine and Biological Sciences, Aldermoor Health Centre, Southampton S016 5ST,
- b Park Surgery, Chandler's Ford, Hampshire SO5 2ZH
- Correspondence to: Dr Little
- Accepted 6 April 1998
Objective: To assess whether provision of educational leaflets or questions on contraception improves knowledge of contraception in women taking the combined contraceptive pill.
Design: Randomisation of women into three groups according to type of educational leaflet on contraceptive information. These groups were subdivided into two on the basis of questions on contraception asked by the doctor or practice nurse. The women were followed up by postal questionnaire 3 months later.
Setting: 15 general practices in South and West region.
Subjects: 636 women attending check up appointment for repeat prescription of the combined contraceptive pill.
Main outcome measures: Knowledge of: factors causing pill failure, subsequent action, emergency contraception, and all the rules (pill rules) that apply to the contraceptive pill.
Results: 523 women returned completed questionnaires (response rate 82%). Knowledge of contraception with no intervention was low with only 10 (12%) women knowing all the pill rules. Educational intervention had a highly significant effect on knowledge of: factors causing pill failure (likelihood ratio χ2=22); subsequent action (21); emergency contraception (24); and all the pill rules (22) (P<0.01 in all cases). Improvement in knowledge of all the pill rules occurred with provision of the summary leaflet (28% knew all the rules, adjusted odds ratio 4.04, 95% confidence interval 1.68 to 9.75), the Family Planning Association's leaflet (27%, 3.43, 1.45 to 8.09), and asking questions (26%, 3.03, 1.30 to 7.00). Asking questions in addition to provision of leaflets improved knowledge of contraception further for the summary leaflet (39%, 6.81, 2.85 to 16.27) but not for the Family Planning Association leaflet (21%, 2.58, 1.07 to 6.18).
Conclusion: Women attending check ups for repeat prescriptions of the contraceptive pill should be provided with educational leaflets on contraception or asked relevant questions to help improve their knowledge of contraception. Asking questions in addition to providing a summary leaflet is time consuming, but results in the most knowledge gained.
It is very important to most women who take the contraceptive pill to avoid pregnancy
Poor knowledge of contraception is common in women taking the contraceptive pill and associated with unwanted pregnancy
No randomised trials have been conducted of educational interventions to improve knowledge of contraception
Providing leaflets or asking questions both improve knowledge of contraception
The largest effect on knowledge of contraception is from asking questions and providing a summary leaflet
Poor knowledge of the combined contraceptive pill is well documented1-5 particularly in women attending appointments in general practice, where most contraceptive pills are prescribed.4 This lack of knowledge may be a major contributing factor to pill failures and to about 20% of unwanted pregnancies. 1 4 6
Whether knowledge of contraception leads to better pill use and less unwanted pregnancies remains to be clarified. However, improving womens' knowledge of the contraceptive pill is essential if unwanted pregnancies are to be avoided; an important factor for most women taking the contraceptive pill.7
Although medical literature does advise providing women with leaflets and repeated verbal counselling on taking the contraceptive pill, 6 8 to date the effectiveness of these interventions is supported only by studies that are uncontrolled or unrandomised. 9 10 The evidence base for such interventions—particularly during check up appointments for repeat prescriptions, which provides an ideal opportunity for giving advice—is very limited.
Our randomised study assessed the effect of two types of leaflet, and of teaching the rules of the contraceptive pill (pill rules), by asking women questions about knowledge on contraception.
Subjects and methods
Setting —We wrote to all practices listed by two health commission areas in the South and West region of England. Overall, 38 doctors and seven practice nurses from 15 general practices agreed to participate. All the practices were teaching facilities, nine were training practices and eight fundholding practices. The mean patient list size was 9183 (SD 2207).
Statistics— Sample size was calculated with EpiInfo, providing 80% power for 95% confidence, and used sample size guidelines for factorial studies.11 Based on previous studies,9 507 women (or 633 allowing for a non-response rate of 20%) would be needed to detect an absolute difference of 10% in those knowing all the pill rules when comparing two types of leaflet with no leaflet. This would allow detection of an 8% difference between asking and not asking questions.
Patients —From June to November 1996, a total of 636 women aged 18-45 attended a check up appointment for repeat prescription of the combined contraceptive pill at the 15 general practices participating in the study. Women were excluded if they were: aged 17 or under, owing to the sensitivity of collecting the postal outcome measures; unable to complete the questionnaire (learning disability, schizophrenia, major current manic or depressive episode, or both); or if the consultation was their first for the contraceptive pill (this created the ethical dilemma of randomising such patients to a control group where they would receive no education on the pill rules).
Ethics —If we had provided the patients with a full discussion of the aims of the study and of the randomisation groups this would have biased the outcomes and prevented a meaningful control group. Furthermore, the control group represented normal practice, where most women are not routinely retaught all the pill rules. We therefore sought and were granted the approval of the local ethics committees to consent women to a study where they would be sent a confidential questionnaire about their contraceptive pill after 3 months.
Assignment —During consultation women were randomised to one of six groups in a 3 × 2 factorial design (fig 1). This was achieved by using sealed opaque envelopes that were numbered. Each envelope contained a simple management sheet for the group, containing items on what to do and boxes to tick once each item had been covered. The management sheet was determined by random numbers tables by one of us (JK).
Interventions —The interactive questions were itemised, and the doctor or practice nurse posed each item to the patient and ticked the box as appropriate. If the women could not respond the correct answer was discussed. We aimed for an interactive rather than a simple didactic approach. The interactive questions took 2-5 minutes depending on the woman's level of knowledge. We used two kinds of leaflet. The first was a laminated leaflet, the size of a credit card, that contained a summary of the pill rules. The card was developed from discussions with women who took the contraceptive pill, doctors, nursing colleagues, and our advisors. The second leaflet was the latest one produced by the Family Planning Association. The leaflets were given to the women without explanation, but with a simple endorsement by the doctor or practice nurse. The management sheets and the answers to the interactive questions were all well completed.
Concealment of allocation —The consent form and intervention material were placed in stiff card in an opaque envelope. The summary leaflets were taped to the card to prevent movement, and a blank leaflet was added for groups not requiring family planning leaflets. In this way the appearance, weight, and feel of each envelope was similar; it was also impossible to guess the intervention group in pilot testing.
Outcome assessment —After 3 months the women were sent a questionnaire based on knowledge of contraception. This questionnaire had been previously validated for face, content, and construct validity.7 As 3 months is the mid point between 6 monthly appointments it should provide a good index of the average improvement in long term knowledge. The women were asked to answer all questions from memory without referral to leaflets. Second and third mailings were sent to non-responders. Data were entered by a research assistant (JK) blind to the randomisation group.
Data were analysed using SPSS (SPSS, Chicago, IL) and Stata (StataCorp, College Station, TX) for Windows software packages. The principal intention to treat analysis of the responders' knowledge used logistic regression for a factorial study to assess the odds ratio of improving knowledge of all 12 basic pill rules (box) adjusted for the important predictors of knowledge. The effect of each variable and of interaction between the two factors was tested using the likelihood ratio χ2 test: terms significant at the 5% level using this test in univariate analysis were included in the logistic model by forward selection to adjust the estimates, starting with the most significant terms. Where interaction was significant we presented the results of the individual groups defined by the two factors.
Twelve basic pill rules
Factors associated with pill failure
Missing pill by 12 hours
Starting packet late
Subsequent action after pill failure
Continue taking the pill
Extra precautions (barrier methods)
Use extra precautions during pill failure and for 7 more days (7 day rule)
Run the packets together if missed pill during the past week
Emergency contraception if had sex in pill free week and starts packet late
72 hour time limit
Knowledge of all the pill rules as an outcome may underestimate the ability of interventions to reduce the risk of pregnancy: any improvement in the number of rules women know is likely to improve the efficacy of pill use in the population. Thus we compared the differences in the number of basic rules that women know—using the Kruskal-Wallis test to compare groups as the data was skewed.
A secondary outcome was the number of women with knowledge (score) at follow up that was greater than the median, using logistic regression. The 32 item score included less essential aspects of knowledge of contraception—for example, factors not causing pill failure, the coil as emergency contraception, timing of pill use—and has good construct validity.7
Five hundred and thirty seven women (84%) returned their questionnaires (responders); 523 (82%) completed all knowledge questions.
Different numbers occurred in each group owing to the variable recruitment of patients by doctors and practice nurses in the practices and the lack of block randomisation (which in retrospect would have been preferable). There was no evidence, however, of selection bias, there was thorough concealment of allocation, and similar characteristics were evident in the responders for the different levels of each factor (table 1), except women who reported use of emergency contraception in the past. This is a chance finding, but as it is a potentially important confounder the odds ratios are adjusted for this variable and for other important predictors of knowledge. The number of women (n=10) knowing all the pill rules with no intervention was low (12%, table 2). For most of the women, avoiding pregnancy was important (table 1).
The number of responders (139/264, 53%) who correctly answered the interactive questions at baseline in at least one of three areas of knowledge (pill failure, subsequent action, or emergency contraception) was not significantly different from non-responders (25/46, 54%) (χ2=0.04, P=0.8).
Selection bias —Doctors and nurses who recruited 20 or more patients were known as high recruiters. There was no significant difference in the characteristics of patients from high recruiters compared with low recruiters for those answering the interactive questions correctly for one area of pill knowledge (61/129 (47%) and 103/181 (57%) respectively) (χ2=2.8, P=0.09), importance of not getting pregnant (177/224 (79%) and 232/312 (74%)) (χ2=1.6, P=0.21), social class I/II (60/157 (38%) and 110/252 (44%)) (χ2=1.2, P=0.27), or education over age 15 (135/222 (61%) and 200/310 (65%)) (χ2=0.8, P=0.38).
Effect of leaflets and asking interactive questions
The interaction between leaflets and asking questions was significant or very close to significance for knowledge of factors causing pill failure (likelihood ratio χ2=5.8, P=0.056), subsequent action (5.07, P=0.08), emergency contraception (5.76, P=0.056), and knowledge of all the pill rules (6.23, P=0.04). Thus the results for the six groups defined by the two factors are presented separately (table 2). The results suggest that all single interventions (both leaflets, and asking questions) produce a modest improvement in knowledge. Additional benefit is conferred only from asking questions with the summary leaflet.
Fig 2 (which excludes outliers) shows that intervention improves knowledge of the basic pill rules at all levels of knowledge (Kruskal-Wallis χ2=33, P<0.001). Asking questions particularly helps women with poor knowledge.
Knowledge score —The odds ratio (95% confidence intervals) of having a high knowledge score at follow up, adjusted for other predictors of knowledge of contraception, was: summary leaflet 1.73 (0.87 to 3.44); Family Planning Association leaflet 3.45 (1.75 to 6.78); interactive questions 2.14 (1.12 to 4.11); summary leaflet plus questions 4.41 (2.17 to 8.97); and Family Planning Association leaflet plus questions 3.31 (1.69 to 6.50).
As far as we are aware, this is the first randomised trial in primary care of strategies to improve knowledge of contraception.
Limitations of the study
Exclusions —The study excluded women aged 17 or under, but included most women on the combined contraceptive pill in whom educational intervention would be considered.
Generalisability —Although only doctors and nurses from training practices participated, other practice characteristics were representative and the interventions are likely to be generalisable as the doctors and nurses were constrained by the management sheets to deliver standard interventions. Furthermore, the patient characteristics are similar to those of the largest uncontrolled study,10 and comparing patients from high and low recruiters showed no obvious selection bias. The groups had similar characteristics, with the exception of one potential confounding variable which was used to adjust the results. These results may not generalise to groups with different levels of knowledge for example, women attending family planning clinics4: separate trials are needed in these settings.
Non-response —A response rate of 82% was achieved, and the knowledge of contraception of the responders was similar to the original randomised group.
Outcome assessment —As knowledge was self reported using a questionnaire there was nothing to stop the women from obtaining information from the inserts in their contraceptive pill packs, books, or the leaflets. To minimise this bias we asked the women not to check their answers. Even if this was the explanation for the results, this suggests intervention at least improves access to knowledge; furthermore, it does not explain the differences between groups and is unlikely to be a significant factor for most women since the final levels of knowledge were still not high. Thus despite potential limitations, the study provides reasonable estimates of the effectiveness of the interventions.
Effect of interventions on knowledge
This study shows the importance for most women of avoiding pregnancy, and confirms previous evidence that many are at risk of pregnancy because of poor knowledge of contraception.1-5 Single interventions modestly increased both the number of pill rules known and the number of women knowing all the pill rules, supporting evidence from unrandomised and uncontrolled trials. 9 10 Leaflets are the most efficient way of increasing the number of women knowing all the pill rules as they only require seconds for a clinician to endorse. Asking questions may particularly improve understanding in those women with very poor knowledge. There is only additional benefit in asking questions when using a summary leaflet that presents information in a similar format to the questions, and not when using the more complicated Family Planning Association leaflet. Simple summary leaflets as laminated cards are not routinely available: if they are, health professionals will need to decide if the 2-5 minutes spent asking the extra questions is worth the knowledge gained.
Improvement in knowledge of contraception is important as most women want to avoid pregnancy, and yet very few know the basic rules for avoiding pregnancy. Health professionals should provide leaflets or ask questions on knowledge of contraception, or both, at consultations for repeat prescriptions of the contraceptive pill, as education seems to improve knowledge of contraception even after one consultation.
We thank the following doctors and practice nurses for their help in recruitment, constructive comments, and enthusiasm: Drs Stephens, Gibson, Barnsley, Bond, Moore, Glaysher, Stanger, Higgins, O'Connor, Podkolinski, Dickson, Terry, Thompson, Wood, Munro, Stobbs, Darch, Boyd, Godfrey, Lupton, Chaplin-Rodgers, Richenbach, Bacon, Fowler, Mooney, Mansell, Snell, McCallum, Meakins, Baber, Ord-Hume, Rees-Jones, Das, and LeBesque, and sisters Redman, Barr, Davies, McCrea, Roe, Bunyan, and Winter. We thank Dr Christine Glew, Professor John Guillebaud, Dr Elaine Cooper, and Dr Sarah Randall, the patients and staff at Aldermoor Health Centre, the general practitioners, and the family planning doctors who gave their time during development of the questionnaire. We thank the Family Planning Association for the use of their leaflets.
Contributors: PL had the original idea for the present study, led the grant application, supervised the development and conduct of the study, performed the main analysis, and wrote the main drafts of the paper; he will act as guarantor for the paper. SG contributed to writing the grant application, supervised the development and conduct of the study, jointly performed the main analysis, and contributed to the writing of the paper. JK managed all phases of the study, entered and checked the data, contributed to the development of the interventions, the questionnaire, the analysis, and writing the paper. ND contributed to the grant application, development and supervision of the study, development of the interventions and the questionnaire, and writing the paper. CS contributed to the development and supervision of the study, and the development of the interventions, the questionnaire, the analysis, and writing the paper. Dr Christine Glew was responsible for the day to day management of the development phase of the study and contributed to the development and validation of the questionnaire. Professor Guillebaud, Dr Sarah Randall, and Dr Elaine Cooper contributed to the development of the questionnaire and the interventions.
Funding: This work was supported by Wessex NHS regional research and development funds. PL and SG are supported by the Wellcome Trust.
Conflict of interest: None.