Sex education and teenage pregnancy

Submission to the Social Exclusion Unit
18 November 1998

Analysis of:

Kirby D. Sex and HIV/AIDS education in schools - Have a modest but important impact on sexual behaviour [editorial],BMJ, volume 311, 12 August 1995;
Mellanby AR, Phelps FA, Crichton NJ, Tripp JH. School sex education: an experimental programme with educational and medical benefit, BMJ, 311, 414-417, 12 August 1995; and
Wellings K, Wadsworth J, Johnson AM, Field J, Whitaker L, Field B. Provision of sex education and early sexual experience: the relation examined. BMJ, 311, 417-420, 12 August 1995.
NHS Centre for Reviews and Dissemination, Preventing and reducing the adverse effects of unintended teenage pregnancies, Effective Health Care, 3, 1; February 1997.

Introduction

Many factors affect the rates of teenage pregnancy, and while the proximate biological causes are easily stated, the more remote social and interpersonal factors, and how these interplay are far from simple.

There is a great desire to identify factors that can be influenced by policy decisions and hence reduce conception rates among adolescents. At the same time, the moral values of policy makers pre-dispose them in favour of some policies and against others.

Those conducting research, and those analysing or surveying it also have their moral and ethical viewpoints which often colour how information is presented, even in academic journals. Here we look at two instances of the presentation of information - one from a medical journal, the other from the NHS Centre for Reviews and Dissemination, an academic service body. Both might be expected to present results impartially, but in both cases research is presented with a definite "spin."

Those charged with developing social policies must analyse carefully such studies, need to look at the methodology employed, the verifiability of the data, and whether the references cited support the arguments made.

British Medical Journal, volume 311, 12 August 1995

Two research reports appear in this edition of the BMJ, preceded by a leading article, which introduces them, and points to a particular interpretation of the results, justifying this in one case with an argument that was actually countered, rather than supported, in the report referred to.

The editorial is headed: "Sex and HIV/AIDS education in schools [h]ave a modest but important impact on sexual behaviour"

The editorial naturally wants to suggest the two studies following are of some significance. While the "modest" scale of the impact of the programmes is undeniable, to call their impact "important" is less easy to justify. The editorial writer (Douglas Kirby) sums up the main conclusion of one of the articles: "having schools as the primary source of sex education might have increased the use of condoms at first intercourse."

We would make three points about this statement:

Figure 5 of the relevant paper (Wellings et al, p.420) indicates NO significant increased likelihood for men using condoms at first intercourse, regardless of what they specified as the "main source" of sex information, whether school or other. It indicates only a modest increase (just above the level of statistical significance) for girls - but this was not as great among girls who said the school was the main source of sex information as among those who said that a relative (presumably the mother or a sister in most cases) was the main source. (A relative was also marginally more likely than school to influence boys to use condoms at first intercourse, but again this was not statistically significant.) But the "spin" is evident: the stress is on school based sex education, not parental or home centred guidance, both in the research paper itself and in the editorial.
Condom use at first intercourse is hardly a very notable influence of a school based sex education in comparison to other findings in Wellings' report. For example, the proportion of men citing school lessons as their main source of sexual information increased from 7.9% among the oldest age group (45-59) to 27.2% of those in the youngest (16-24). While the proportion of men reporting first intercourse before 16 increased from 10.0% in the oldest group, to 25.4% for the youngest. The authors do not draw this comparison directly, though they do note that "the school has been playing an increasingly important role in the sexual education of the young, particularly for men." Of course, the nature and content of school sex education has changed so drastically that this comparison is not terribly helpful, but it should perhaps suggest that a rethink of the policy is needed.
A further point to note is that no estimate is given of the effect, if any, of condoms at first intercourse on the risks of 'conceiving as a teenager', AIDS/HIV infection, other sexually transmitted diseases, cervical cancer, etc.

A deeper problem with the Wellings study, which the authors acknowledge, is that those who start intercourse early are not so likely to be influenced by school sex education which may post-date sexual initiation. So those who say they got most sexual information from school and used condoms when first having sex may well have been acting under a different influence altogether.

This point may account for the negative conclusion of the paper - that it provides no evidence to support the concern that school sex education might hasten the onset of intercourse. But more importantly, the article can hardly sustain Douglas Kirby's 'spin statement' in the editorial, quoted above, that: "having schools as the primary source of sex education might have increased the use of condoms at first intercourse."

The other study that Kirby's editorial introduces in the BMJ is by Mellanby et al (p.414). Of this he says that it shows that the Mellanby programme delays the onset of sexual activity, in comparison to sex education that the control groups were receiving, which is not described.

While knowledge of contraception was promoted in the programme analysed in this study, it used various strategies aimed directly at decreasing sexual behaviour, something which many advocates of contraceptive-focused school sex education programmes claim is virtually impossible. And being difficult (or perhaps ideologically unattractive to them) they are not willing to try it.

It is clear that certain moral principles were adhered to by those conducting the study: they did not use the potentially powerful technique of stigmatising sexual activity - believing this to be intolerant, and increasing tolerance was another objective of the programme. It is interesting to note that moral values also appeared to influence the behaviour of men and women in Wellings' study, with both men and women who were Roman Catholics less likely than those of other religions to use condoms and other contraceptives at first intercourse than other groups.

Negotiation in relationships and assertiveness were among the social skills promoted in the study, which it would seem fair to say in this case produced a modest impact.

Mellanby recognises that: "Neither specific teaching about contraception nor improving the contraceptive service consistently increase effective contraceptive use by teenagers." (p.414)

The remainder of Kirby's editorial stresses that, when one looks at sex education/AIDS prevention studies all together they don't have an effect in hastening first intercourse. It would seem to be implicit in this that some do hasten intercourse and others retard it, but the stress is on the 'neutrality of programmes overall.' This surely is not good enough. Those programmes that have a positive outcome must be promoted and emulated. Kirby goes on to point out nine factors of effective programmes. These include proscriptive elements such as:"a narrow focus on reducing specific risky behaviour" and "clear reinforcement of individual values and group norms" (i.e. morality)

Interestingly, contraceptives as such are not among the nine key factors for success. The list does include: "activities to convey the risks of unprotected sex and how to avoid them" - a phrase which could equally cover abstinence or "how to say 'no'" strategies, as 'where to go to get your preferred contraceptive.'

While the "spin" put on the results of these articles is deeply regrettable, the following example shows an even more disturbing bias.

Preventing and reducing the adverse effects of unintended teenage pregnancies, NHS Centre of Reviews and Dissemination, University of York, Feb 1997, 3,1.

(Referred to below as "NCRD")

This paper is presented in the format of a number of "headline" claims, followed by an analysis of published studies and reports.

Several of these "headlines" are either misleading or unsupported by the data cited in the document. For instance:

NCRD Headline: "Increasing the availability of contraceptive clinic services for young people is associated with reduced pregnancy rates." In contrast to the positive assertion of this headline, NCRD states:

"The literature searches revealed a complete lack of UK-based controlled evaluations of the effectiveness or cost-effectiveness of different approaches to the delivery of contraceptive services to young people."(p.7) (sic)
"Evaluation of school-based clinics providing health and contraceptive services in the USA have been methodologically weak with poor selection of comparison groups and the results are contradictory. Some show delay in sexual initiation, and reduction in birth rate..." (p.4) [... and presumably the 'contradictory' ones show the opposite.]

NCRD Headline: "Contraceptive services should be based on an assessment of local needs and ensure accessibility and confidentiality" Evidence to support this within the report is poor: "Studies show an association between conception rates and the level and type of contraceptive services available locally." (p.7)

The only reference cited in support of it is an article entitled "Adolescent premarital childbearing: do economic incentives matter?" (Lundberg S., Plotnick RD, in the Journal of Labour Economics (J Lab Econ) 1995, 13, 177-200). As this is the only source given, and the text refers to "studies" in the plural, it is presumably a secondary source and as the statement does not specify whether all the studies found a negative association it gives little support to the assertion in the headline. It would seem to defeat the object of this kind of publication to use a secondary source, especially as this is not only a key recommendation but also one which is controversial for many.

"There is not the sort of clear evidence of the effectiveness of different approaches to contraceptive counselling and contraceptive provision which could provide a firm basis for decision making." (p.7)

"A review of descriptive UK studies [as opposed to controlled evaluations] which examined factors likely to influence the effectiveness of services was carried out." (p.7)

The authors go on to admit the weakness of descriptive studies, but nevertheless, they are sufficiently confident of the review cited to base recommendations on it. The review in question was conducted by the NCRD (the source of the current review), publication "forthcoming." This would seem poor academic practice.

"In order to attract young people to use services, they need to be well-advertised, easily accessed outside school-hours (opening times and location), informal and for under 16-year-olds, confidential." (p.7)

The reference cited in support of this statement is not the product of academic researchers, nor even a descriptive study; it is a paper issued by BMA and other medical bodies in association with the Brook Advisory Centres and the Family Planning Association - organisations committed to birth control programmes.

It should be noted that the factors mentioned (advertising, access, informality, etc) are said to be required "in order to attract young people" - and there appears to be no claim that trying to draw teenagers to use such services will result in any delay in first intercourse or reduction in pregnancy or abortion rates.

NCRD Headline: "School-based sex education can be effective in reducing teenage pregnancy especially when linked to access to contraceptive services." In support of this contention, NCRD summarises its assessment of sex education programmes under seven headings:

Abstinence programmes;
School-based skills building combined with factual information;
Programmes encouraging vocational development;
School-based programmes linked with contraceptive services;
School-based and school-linked clinics;
One-to-one counselling;
Non school-based educational programmes in HIV prevention

(This categorisation is somewhat arbitrary but this is not the focus of our concerns.)

The headline would appear to suggest that protocols 4 and 5 were the successful ones. This is not reflected in the document however.

Of number 4, it is said that such programmes increase the use of contraceptives - but this is not the same as reducing teenage pregnancies (this is a highly controversial point - indeed, one randomised controlled trial referred to elsewhere in the NCRD study (p.5) found that a peer-led programme promoting decision making skills resulted reduced contraceptive use and reduced pregnancy rates (but not in this case by a significant level - at a 95% confidence interval). Another study is reported to have found a significant reduction of pregnancies by combining "peer-led skills and confidence building programmes and access to condoms and transport to contraception clinics." (p.4) Unfortunately the details of this study are not reported. It is not clear whether the reduction was great or small, or even whether it was significant at the 95% confidence interval level. It is not tabulated in the NCRD's table, so presumably it was not a randomised controlled trial.

Four studies are referred to under protocol number 5, but again, none of these appears to have been an RCT. Their results are described by NCRD as "contradictory".

In contrast, the effectiveness of protocols 1, 2 and 3 are nowhere reflected in the NCRD headline recommendations, although there appears to be more evidence for their success.

Protocol 2 includes a reference to the Mellanby study, discussed above. As pointed out, this did show a modest impact on sexual behaviour, in terms of delaying first intercourse (control population odds ratio 1.45 for having had intercourse at time of assessment (15.5-16.5 years of age), compared to intervention group, 95% confidence interval: 1.13-1.87). The impact of the programme appears to have been much stronger on boys than on girls, but separate O.R.s are not quoted by sex.

The NCRD suggest that another programme which did not, like Mellanby's, include information on contraception, was less effective. But they only include data for girls, so it is not possible to judge whether the comparison is valid.

Protocol 3 was apparently fairly effective, one programme having shown a reduction in pregnancy rate. Results are included from a study that reported on a controlled trial (rather than a randomised controlled trial). The controlled trial involving >5,000 pupils, showed a significant reduction in pregnancy rates (odds ratio: 0.39, 95% confidence interval: 0.2, 0.78). Not all of the RCT studies reported included figures for pregnancy rates, but none of those that did showed a reduction significant at the 95% confidence interval. In the light of this, the question must be asked as to why NCRD played down this kind of approach.

NCRD further criticises protocol 3 by saying that it is difficult to separate the contribution of the various elements of the programmes. This seems a very weak point to make. Furthermore there is no suggestion that the different elements of the favoured protocols - 4 and 5 - have been or could be analysed for their separate elements.

Of protocol 1 (described as 'abstinence programmes') NCRD says: "When compared to the usual sex education, abstinence programmes were not found to have any additional effect on either delaying sexual activity or reducing pregnancy." (p.3)

Two references are cited in support: one is by Jorgensen. (Jorgensen, SR. Project taking charge; an evaluation of an adolescent pregnancy prevention program, FAMR 1991;40, 373-80)

The Jorgensen study (and a six month follow-up - apparently mis-quoted as a six week follow-up) is summarised elsewhere in the NCRD study (p.5). The summary contradicts this statement. The study did apparently indicate a lower rate of participants initiating intercourse (and a much lower rate among boys) in an "abstinence programme". This result is dismissed, presumably because, as a small study, it did not achieve a result significant at the 95% confidence interval. The lack of statistical power in the study must be noted, but to claim in the light of this that such programmes were not found to have "any additional effect" is untrue. (The control for the study is referred to as "usual instruction"). The other study cited is not summarised in NCRD (presumably because it was not a randomised controlled trial), and it is not available to us at present, but it would be important to examine it.

Overall, one must regard as highly misleading the headline claim that "School-based sex education can be effective in reducing teenage pregnancy especially when linked to access to contraceptive services." Nearly all of the RCT studies presented in summary form in the document show at most an increase in contraceptive uptake from such programmes, not a decrease in pregnancy rates, as NCRD claims. Even the programme which has the most marked increase in contraceptive uptake (Baker C. Self-efficacy training: its impact upon contraception and depression among a sample of urban adolescent females [PhD], Seton Hall University, 1990) failed to achieve a statistically significant reduction in pregnancy rates at the 95% confidence level.

A final point to be noted about this study is that among those listed as assisting the study and commenting on the text are Alison Hadley of the Brook Advisory Centre, Joan Walsh of the Family Planning Association, and Caroline Ray of the Sex Education Forum. These organisations are strongly ideologically committed to the provision of contraception to teenagers. If impartiality on a controversial topic was desired, one would have expected bodies critical of the proposed policies to be canvassed as well, or for those with such evident axes to grind to be excluded.