Family Practice Vol. 18, No. 4, 449-453
© Oxford University Press 2001
Unwanted pregnancy and contraceptive knowledge: identifying vulnerable groups from a randomized controlled trial of educational interventions
Primary Medical Care Group, University of Southampton, Aldermoor Health Centre, Aldermoor Close, Southampton SO16 5ST and
a Park Surgery, Chandler's Ford, Southampton, UK.
Little P, Griffin S, Dickson N and Sadler C. Unwanted pregnancy and contraceptive knowledge: identifying vulnerable groups from a randomized controlled trial of educational interventions. Family Practice 2001; 18: 449–453.
Received 8 December 2000; Accepted 12 March 2001.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Objectives. The aim of this study was to identify predictors of contraceptive pill knowledge and their relationship to educational interventions.
Methods. A total of 636 women attending for a follow-up appointment for repeat prescription of the combined oral contraceptive pill with a GP or practice nurse were randomized to receive leaflets (simple summary leaflet or FPA leaflet), advice or neither. Sociodemographic details and contraceptive knowledge were determined using a validated contraceptive knowledge questionnaire sent after 3 months by post. The main outcomes were sociodemographic, contraceptive, attitudinal and educational predictors of knowledge.
Results. A total of 522 (82%) had complete questionnaires. After controlling for educational intervention and other confounding variables, independent predictors of knowledge were further education (adjusted odds ratio 2.98, 95% confidence interval 1.78–4.99); number of years on the pill (0–5, 6–10, >10 years) 1.0, 0.56 (0.33–0.95) and 0.34 (0.19–0.59), respectively; past emergency contraception (1.87, 1.18–2.97); and importance attached to not falling pregnant (1.83, 1.02–3.29). These predictors are less powerful than the impact of most educational interventions (range of odds ratios for interventions: 1.85–6.81), and there was no evidence of a separate effect of educational intervention in any subgroup, except that leaflets have a larger effect in women who have needed emergency contraception in the past (no past use or simple summary and FPA leaflets, 1.74 and 0.90, respectively; with past use, 3.47 and 3.83; interaction term chi-square 6.92, P = 0.03).
Conclusion. Educational interventions are as important as sociodemographic features in determining knowledge. With limited time for full educational interventions in practice, priorities for intervention should be women who have used emergency contraception in the past—who will benefit most—and those on the pill for >5 years or with no further education who are at highest risk due to poor knowledge.
Keywords. Contraception, education, knowledge, leaflet.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
An estimated 3 million women (or 23% of all women using contraception) use the combined oral contraceptive pill (COCP).1 Poor knowledge of effective use of the pill and of secondary contraception is well documented,2–6 particularly in general practice,2,4 leading to pill failures2 and unwanted pregnancy.5,7 Thus poor contraceptive knowledge is an important area for intervention in primary care since contraceptive knowledge is one of the few modifiable risk factors for unwanted pregnancy, and most women taking the COCP are managed in primary care.
GPs have been advised to counsel women repeatedly about pill use and what to do in the event of failure, using both written information and verbal advice.7,8 An uncontrolled trial9 and a recent randomized controlled trial demonstrated that educational interventions can improve knowledge.10 However, since the social context may be crucial, and the realities of everyday practice mean that GPs have limited time to perform full educational interventions,10 important questions remain unanswered. These questions include: which of sociodemographic, educational or clinical variables are most important in determining knowledge; which women are particularly at risk; and do interventions work equally well in all groups? Knowledge of which factors predict contraceptive knowledge and how these variables relate to educational interventions is a priority since it will allow appropriate targeting of interventions in primary care.
To date, there have been no randomized trials where the effect of both intervention and other important predictors of knowledge have been assessed simultaneously. The largest observational study from primary care to date has provided the most useful information to identify vulnerable groups.9 However, this study was limited in a number of ways: (i) it concentrated on factors predicting emergency contraception, and did not assess other aspects of basic knowledge, thus limiting the ability to identify vulnerable groups; (ii) it did not assess potentially important predictors of knowledge such as patterns of current pill use and of attitude to falling pregnant; and (iii) it was an uncontrolled study and thus could not assess interaction between intervention and sociodemographic variables or predictors of knowledge9 to identify groups that will particularly benefit (or disbenefit) from intervention.
We report an analysis from a large representative trial cohort,10 which is clinically important to inform intervention in priority groups and furthers our understanding of the epidemiology of a modifiable risk factor for unwanted pregnancy.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Full details of the trial cohort and validity of the questionnaire have been reported:10,11 the trial data set is not likely to be limited by selection bias, non-response or outcome assessment bias.10 A total of 636 women were randomized at repeat pill checks to one of three leaflet groups [no leaflet, credit card sized summary of pill rules (summary leaflet) or family planning association leaflet] and two question groups (no questions or asking structured questions about contraceptive knowledge). Patients were sent a validated contraceptive knowledge questionnaire by post after 3 months.11 The questionnaire also asked for sociodemograpic details and details of current and past contraceptive use. Eighty-two per cent of questionnaires were returned fully completed. Patients were not given the questionnaire at baseline since the questionnaire itself is potentially an ‘intervention’, and it plausibly could also artificially improve response to the leaflets or advice. Three months was chosen as a reasonable marker of long-term knowledge gain, and was half way between 6-monthly pill checks.11
Sample size (for 80% power and 95% confidence)
Assuming 10% of ‘unexposed’ individuals had good knowledge,2–6,9–11 520 individuals were needed to detect a 10% absolute difference in knowledge in exposed individuals, assuming exposure prevalences of 30–65%.
Analysis
To assess factors predicting the key outcome (good knowledge, i.e. knowing the basic pill rules;10 see footnote to Table 1
), terms were included in a logistic regression model by forward selection to adjust the estimates, starting with the most significant terms first, and retained if they remained significant (5% level).
|
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Important predictors of good knowledge were further education, shorter duration of pill use, attitude to pregnancy and educational interventions (Tables 1 and 2
). To assess whether the effect of intervention on knowledge of the rules is sensitive to sociodemographic features of the study population, or to important predictors of knowledge, interaction was assessed for both the leaflet and questions with educational level [likelihood ratio (LRT) chi-square for interaction terms 0.1, P = 0.9, and 0.3, P = 0.6, respectively], importance of not getting pregnant (LRT chi-square 2.1, P = 0.3, and 0.11, P = 0.7, respectively), social class (LRT chi-square 0.6, P = 0.7, and 0.3, P = 0.6, respectively), age (LRT chi-square 1.0, P = 0.9, and 2.0, P = 0.4, respectively) and time on the pill (LRT chi-square 5.5, P = 0.23, and 1.1, P = 0.6, respectively). The only significant interaction was between the effect of leaflets and past use of emergency contraception: there was a significantly larger effect of both types of leaflet (odds ratios for simple summary and FPA leaflet alone 1.74 and 0.90, respectively; with past use of emergency contraception 3.47 and 3.83, respectively; LRT for interaction term 6.92, P = 0.03)—which explains much of the increase in knowledge with past use of emergency contraception.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
This is the largest study, and the only randomized trial to report the predictors of knowledge and the effect of simple educational interventions for adult pill users in general practice. Before interpreting the results, the limitations of this study will be discussed.
Study design
What are the potential limitations of the study? Younger teenagers were excluded due to the difficulty of sending postal questionnaires about sensitive subjects to family environments. Nevertheless, the trial addresses the large group where GPs and practice nurses might consider providing information/teaching about pill rules. Selective overall recruitment may limit generalizeability. However, the sociodemographic characteristics of our study are similar to those of the largest observational study,9 comparing patients from high and low recruiters showed no obvious bias,10 and the results were not sensitive to sociodemographic features or to other important predictors of knowledge. Variable or low response rates can jeopardize generalizeability; however, a >80% response was achieved, and the ‘responders’ were similar in terms of contraceptive knowledge to the original group. Thus the study is likely to provide reasonable estimates of the likely important predictors of knowledge among pill users in primary care.
What factors predict contraceptive knowledge?
Important independent predictors of knowledge supporting previous observational work9 are further education and previous use of emergency contraception. This study also shows that other important independent predictors are the number of years on the pill, and the importance attached to not falling pregnant. In addition, women who start their packet on the same day of the week are more likely to have greater knowledge. The number of years taking the pill is a particularly important clinical point since it seems counter-intuitive: it is easy to assume that women well established in taking the pill would be in no need of being re-taught pill rules. Assuming the standard of teaching of pill rules to women has not changed over the last 10 years, these results suggest that lack of reinforcement of learning is the likely explanation for poor knowledge: thus it is important to provide reinforcement of initial teaching to established pill users.
The effect of interventions, and interaction between interventions and other predictors of knowledge
Educational interventions were at least as powerful as most sociodemographic predictors of knowledge and mostly more powerful. Leaflets are clearly the most efficient way of improving pill knowledge, since they only require seconds to endorse and deliver. The absolute gains in knowledge are modest and in line with previous evidence from unrandomized trials,12 and suggest that the evidence from uncontrolled studies may possibly overestimate the possible gain from using leaflets or reteaching pill rules.9 Most sociodemographic or contraceptive variables did not predict a different response to intervention (i.e. no evidence of interaction), which supports the efficacy of educational interventions in all sociodemographic groups—and the importance particularly in those demographic groups at risk due to poor knowledge. The only evidence for interaction was with women who had needed emergency contraception in the past, in which case leaflets had a significantly greater effect, i.e. this group provide a particularly important opportunity to improve knowledge. Thus this study helps to identify priority groups for intervention: priority groups should be those women with particularly poor knowledge or where the evidence suggests that interventions are especially effective.
|
Conclusion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
Educational interventions are as important as sociodemographic features in determining knowledge. Clinicians should not assume that educational interventions are ineffective among individuals from disadvantaged backgrounds, or among established pill users. With limited time for full educational interventions in practice, priorities for intervention should be women who have used emergency contraception in the past—who will benefit most—and those on the pill for >5 years or with no further education, who are at highest risk due to poor knowledge.
|
Acknowledgments |
|---|
We are very grateful to the following GPs 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; Sisters Redman, Barr, Davies, McCrea, Roe, Bunyan and Winter. We thank Christine Glew, Professor John Guillebaud, Dr Elaine Cooper and Dr Sarah Randall, the patients and staff at Aldermoor Health Centre, the GPs 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. This work was supported by Wessex NHS Regional R+D funds. PL and SG were supported by the Wellcome Trust; PL is supported by the Medical Research Council.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionConclusionReferences |
|---|
1 Anonymous. General Household Survey. London: HMSO, 1986.
2 Duncan G, Harper C, Ashwell E, Mant D, Buchan H, Jones L. Termination of pregnancy: lessons for prevention. Br J Fam Plann 1990; 15: 112–117.
3 Finlay IG, Scott MGB. Pattern of contraceptive pill taking in an inner city practice. Br Med J 1986; 293: 601–602.
4 Brook SJ, Smith C. Do combined oral contraceptive pill users know how to take their pill properly? Br J Fam Plann 1991; 17: 18–20.
5 Bromham DR, Cartmill RSV. Knowledge and use of secondary contraception among patients requesting termination of pregnancy. Br Med J 1993; 306: 556–557.
6 Wilson S, Taylor R. Women's knowledge about contraception. Br J Fam Plann 1992; 18: 3–5.
7 International Working Group on Enhancing Patient Compliance and Oral Contraceptive Efficacy. A consensus statement: enhancing patient compliance and oral contraceptive efficacy. Br J Fam Plann 1993; 18: 126–129.
8 Anonymous. Hormonal contraception. MeReC Bull 1996; 7: 33–36.
9 Smith L, Whitfield M. Women's knowledge of taking the oral contraceptive pill correctly and of emergency contraception: effect of providing information leaflets in general practice. Br J Gen Pract 1995; 45: 409–415.[Web of Science][Medline]
10
Little P, Griffin S, Kelly J, Dickson N, Sadler C. Effect of educational leaflets and questions on knowledge of contraception in women taking the combined oral contraceptive pill: randomised controlled trial. Br Med J 1998; 316: 1948–1952.
11 Little PS, Glew C, Kelly J, Griffin S, Dickson N, Sadler C. Contraceptive knowledge: development of a valid measure and survey of pill users. Br J Fam Plann 1998; 24: 98–100.[Medline]
12 Metson D, Kassianos G, Moriarty J, Norman D. Effect of information leaflets on long term recall—useful or useless? Br J Fam Plann 1992; 17: 21–23.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||