Edinburgh Research Explorer Changes in conceptions in women younger than 18 years and the circumstances of young mothers in England in 2000–12: an observational study

,


Introduction
At the end of the 20th century, concern for high rates of teenage conception in the UK compared with other western European countries, together with the strong association between early parenthood and deprivation, provided the impetus to public health eff orts to prevent teenage pregnancy.In 1999, the UK Government launched a 10-year, nationwide Teenage Pregnancy Strategy 1 in England with the dual aims of achieving a 50% reduction in conception rates in women younger than 18 years by 2010, and mitigating social exclusion in teenage parents by increasing their participation in education, employment, or training. 2][5][6] The teenage pregnancy strategy gave rise to a multicomponent programme comprising a national media campaign; improvements to sex education and young people's sexual health services; support for young parents to increase participation in education, training, and employment; and joint action to ensure national and local coordination across statutory and voluntary agencies. 7Financial resources were allocated according to teenage pregnancy rates in each top-tier local authority area.Because of the concentration of high rates of teenage pregnancy in poorer areas of the country, fi nancial resources were disproportionately invested in the more deprived areas of England.Scotland, Wales, and Northern Ireland decided their initiatives independently.
Conception rates in women younger than 18 years fell by 51% between 1998 and 2014 in England.Of interest is the extent to which the decline seems to be related to the Teenage Pregnancy Strategy and other interventions; whether the decline has been seen equally in the most and least disadvantaged women; and what factors remain associated with conception in women younger than 18 years.In this study, we combined analyses of routinely available area-level data for conceptions, deprivation, and policy-related expenditure, and individual-level data from three decennial waves of the National Survey of Sexual Attitudes and Lifestyles (Natsal) to describe change in outcomes relating to key goals of the English Teenage Pregnancy Strategy-ie, conception rates in women younger than 18 years and the prevalence of participation in education, work, and training in women with a child conceived before the age of 18 years.

Study design and participants
In this observational study, we analysed routine data for births and abortions for top-tier local authorities (top-tier local authorities consist of London boroughs, metropolitan borough councils, and unitary authorities, and are a level of local government with responsibilities including education and social services) in England from 1994 to 2013 and individual-level data from three waves of the National Survey of Sexual Attitudes and Lifestyles (wave 1: 1990-91 [Natsal-1], wave 2: 1999-2001 [Natsal-2], and wave 3: 2010-12 [Natsal-3]).We obtained data by calendar year for the resident population and births and abortions by age of mother from the Offi ce for National Statistics for each of the top-tier local authorities in England.Because of their small resident populations, two authorities (City of London and the Scilly Isles) were combined with more populous neighbouring authorities (Hackney and Cornwall, respectively), resulting in a total of 148 local authorities for analysis.
We also obtained data from the Department for Communities and Local Government at the top-tier local authority level for the 2004 Index of Multiple Deprivation, a score of socioeconomic deprivation based on a weighted average of 38 separate indicators across seven distinct domains (employment, income, health and disability, education skills and training, barriers to housing and other services, crime, and living environment). 8s an indicator of the extent of Teenage Pregnancy Strategy-related local activity, we obtained from the Department for Children, Schools and Families the Teenage Pregnancy Strategy annual Local Implementation Grant awarded to each top-tier local authority for the fi nancial year(s) 1999-2000 and 2010-11, the amount refl ecting the challenge in terms of conception rates in women younger than 18 years and the size of the population.From these data we calculated the total investment per head by dividing the amount by the 2001 Offi ce of National Statistics estimate of resident women

Research in context
Evidence before this study We searched PubMed for studies published between Jan 1, 2000, and March 31, 2016, with no language restrictions using the search terms "teenage pregnancy", "teenage conception", "trends", and "public health policy" to identify studies describing trends in teenage pregnancy and policy-related factors associated with change, in the UK and elsewhere.We examined the titles and abstracts identifi ed by the search, excluding those that were not relevant, and reviewed the full text of the remaining articles to assess appropriateness for inclusion.
Few studies were able to measure direct indicators of interventionrelated action, and none combined general population survey data and routinely collected statistics.Most studies were from the USA and the UK, settings in which the prevalence of teenage pregnancy is high compared with other high-income countries.Existing studies have near universally reported strong associations between higher rates of teenage pregnancy and lower socioeconomic level.They have, however, diff ered in their interpretation of whether the decline in the prevalence of teenage pregnancy in the 21st century can be attributed to broader societal changes, such as increased educational attainment, or to public health interventions such as increased use of reliable contraception.

Added value of this study
Our study aimed to assess progress towards the goals of the Teenage Pregnancy Strategy mounted in England to reduce conception rates in women younger than 18 years, combining analyses of data from a population-based survey and from routinely collected national statistics.The scale of the decline in conception rates in women younger than 18 years, and its association with intervention-related investment and with both demographic and behavioural factors, suggest a combined infl uence of both public health intervention and secular trends on the decline in conceptions in women younger than 18 years in England.

Implications of all the available evidence
The trend towards postponement of key life events such as completion of education, leaving home, starting employment, and settling with a partner is now near universal.These factors seem to have contributed to a global trend towards fewer early pregnancies.Coincidental with this, focused and sustained eff orts to lower the prevalence of early pregnancy by raising awareness, changing social norms, and increasing access to education and reliable contraception seem to have accelerated the trend towards fewer teenage pregnancies.aged 13-17 years.We obtained ethics approval for Natsal-2 from University College Hospital, North Thames Multicentre, and all local research ethics committees in Britain, and for Natsal-3 from the Oxford Research Ethics Committee A.
In Natsal-1, paper questionnaires were self-completed and interviewer administered.In Natsal-2 and Natsal-3, participants were interviewed with a combination of computer-assisted personal and computer-assisted selfinterviews.Experimental comparison of pencil and paper and computer-assisted interviewing revealed no important diff erences in responses. 9For all three surveys, after correcting for unequal selection probabilities, a non-response post-stratifi cation weight was applied to ensure comparability with census data in terms of age, sex, and region.Further details of the methods and response calculations are described elsewhere. 10,11atsal-1 and Natsal-2 asked about the number and timing of all abortions and livebirths; Natsal-3 asked for timing and outcome of each pregnancy.We calculated conceptions that occurred before age 18 years among 18-24-year-old women by summing births that occurred before age 18 years 9 months and abortions occurring before age 18 years, excluding miscarriages and stillbirths, for consistency with Offi ce of National Statistics procedures.We assessed change in the prevalence of participation in any of education, work, or training in young mothers.
4][5][6] They include age at fi rst heterosexual intercourse; use of a reliable method of contraception at fi rst sex; consensuality in terms of being equally or more willing than the partner at time of fi rst sex; autonomy of decision making-ie, not infl uenced by peer pressure or use of drugs or alcohol; and retrospective views on the timing of fi rst sex-ie, whether it occurred sooner or later than ideal or at the right time.We also included the main source of information about sexual matters at the start of sexual activity, adequacy of that information, and ease of communication with parents about sex.Demographic measures included age at interview, family structure (whether the participant lived with both parents until age 16 years [Natsal-2] or age 14 years [Natsal-3]), educational attainment, and area-level deprivation measured with the Index of Multiple Deprivation. 12Full postcode data for use in estimating the Index of Multiple Deprivation were not available for Natsal-1.

Statistical analysis
Analyses using routinely collected data included tabulation and graphical representation of trends in conception rates of women younger than 18 years by quartiles of per head total Local Implementation Grant investment and quartiles of area-level deprivation.Random-eff ects metaregression analyses of the change in conception rates of women younger than 18 years between 1994-98 (the preintervention baseline) and 2009-13 (the most recently available Offi ce of National Statistics data) were based on aggregate top-tier local authority level data.Because of variability in how each top-tier local authority spent their grant, a random-eff ects meta-regression analysis was used.Absolute and percentage changes between these timepoints were analysed as outcomes separately and data from each top-tier local authority were weighted by the inverse of the variance of the outcome estimate.The explanatory factors considered were per head Teenage Pregnancy Strategy investment, region, and deprivation (2004 Index of Multiple Deprivation scores).For the two continuous covariates (Teenage Pregnancy Strategy investment and Index of Multiple Deprivation score), the linearity of their association with the outcome variable (change in conception rate in women younger than 18 years) was assessed by also including quadratic terms in the regression model.
Analyses using Natsal data included logistic regression to examine factors associated with conception in women younger than 18 years separately in 1999-2001 and 2010-12 in women aged 18-24 years, with and without adjustment for the eff ect of all variables except age at fi rst intercourse.Logistic regression with interaction terms was also used to assess whether associations between factors and conception in women younger than 18 years had changed between 1999-2001 and 2010-12.Analyses using Natsal data also included bivariate tabulations and logistic regression to examine the prevalence in 1990-91, 1999-2001, and 2010-12 of participation in work, education, or training in women aged 18-24 years, according to whether or not they had a child conceived before age 18 years.Logistic regression with interaction terms was used to assess whether the association between early motherhood and participation changed over time.All analyses were done with Stata 12.1 (version 13; StataCorp LP, College Station, Texas), accounting for stratifi cation, clustering, and weighting of the Natsal data.

Role of the funding source
The funders had no role in study design, data collection, data analysis, data interpretation, or writing of the report.The corresponding author had full access to all the data in the study and had fi nal responsibility for the decision to submit for publication.

Results
Routinely collected national data for conceptions in women younger than 18 years show a steady decline from their peak in 1998, with an apparent acceleration in that decline from 2007 onwards (fi gure 1).The trend was initially driven by the decline in maternities, until 2007 when the plot lines for abortions and births converged.
The annual rates of conception in women younger than 18 years by quartile of area-related deprivation between 1994 and 2013 declined across all deprivation levels but this decline was larger in the most deprived areas (fi gure 2).Between 1998 and 2013, for example, conception rate declined by 16 conceptions per 1000 women aged 15-17 years in the least deprived areas compared with 33 conceptions per 1000 women aged 15-17 years living in the most deprived areas.This diff erential decline has resulted in a partial convergence in conception rates in women younger than 18 years across the quartiles of deprivation.
The annual data for the same period by quartile of strategy-related expenditure at local level-ie, per head total Local Implementation Grant investment-showed partial gradual convergence in conception rates in women younger than 18 years across Local Implementation Grant quartiles, with the greatest decline in areas receiving the highest Local Imple mentation Grant award (fi gure 2).Between 1998 and 2013, areas receiving the highest Local Implementation Grant award had a decline of 34 conceptions per 1000 women aged 15-17 years, whereas in areas receiving the lowest level, the decline was 16 conceptions per 1000 women aged 15-17 years.
Regression analyses assessing the association between Teenage Pregnancy Strategy funding and decline in conception rates in women younger than 18 years showed an estimated reduction in the conception rate of 11•4 (95% CI 9•6-13•2; p<0•0001) per 1000 women aged 15-17 years for every £100 Teenage Pregnancy Strategy spend per head (table 1).After adjustment for socioeconomic deprivation and region, this trend retained statistical signifi cance, with a reduction of 8•2 conceptions (5•8-10•5; p<0•0001) per 1000 women aged 15-17 years per £100 Teenage Pregnancy Strategy spend per head.This trend was also refl ected when modelling the percentage change in conception rates in women younger than 18 years as the outcome; for every £100 Teenage Pregnancy Strategy spend per head, a 6•2% (95% CI 2•3-10•2) reduction in the conception rate in women younger than 18 years was reported after adjusting for region and deprivation.Testing of quadratic terms indicated that associations with Teenage Pregnancy Strategy spend per head (and Index of Multiple Deprivation score) were approximately linear over the range of values considered for both absolute and percentage change, although the two models are not strictly compatible.The value of I² for both metaregression analyses was almost 100% because the outcome estimates from the top-tier local authorities are all very precise as they are based on rates from tens of thousands of person-years, and hence the variability seen is interpreted as variability in the underlying change across top-tier local authorities.
The prevalence and odds ratios (ORs) of conception in women younger than 18 years by selected characteristics of women aged 18-24 years in 1999-2001 (Natsal-2) and 2010-12 (Natsal-3) are shown (table 2).Comparisons between the two surveys must be cautiously made because of minor but possibly important diff erences in question formulation.However, with this proviso and within the limits of precision of the estimates for the 18-24-year-old age group, the results for the two survey waves are broadly similar with regard to the prevalence of behaviour likely to be proximally and causally associated with conception.The proportion with more than minimum academic   qualifi cations was, however, appreciably higher in 2010-12 (Natsal-3) compared with 1999-2001 (Natsal-2), as was the proportion reporting school lessons as their main source of sex education.A more modest increase between the two timepoints in the proportion reporting use of reliable contraception at fi rst intercourse was noted.
Lower socioeconomic status, lower educational attainment, earlier onset of sexual activity, receiving sex education from sources other than school and, more weakly, negative opinion about the timing of fi rst intercourse were associated with higher conception in women younger than 18 years in both surveys.Interaction testing revealed little clear evidence of change in the association with conception in women younger than 18 years between surveys for any of the factors considered (all p values were greater than 0•05 and are not shown).However, the association between living in an area in the highest quintile of deprivation and conception in women younger than 18  ).The weaker association between use of reliable contraception at fi rst intercourse and conception in women younger than 18 years was further attenuated between Natsal-2 and Natsal-3.
Results from Natsal-1, Natsal-2, and Natsal-3 relating to the participation of women aged 18-24 years according to whether or not they had conceived a child before age 18 years are shown (table 3).Estimates should be treated with caution because of the comparatively small survey subsamples in the 18-24-year-old age group.The results for all three surveys show the proportion of women in education, work, or training at the time of interview was higher by a considerable order of magnitude in those who did not conceive a child before 18 years than in those who did (table 3).However, although the likelihood of participation of women who were not young mothers was unchanged across the three surveys

Discussion
We report a marked decline in conceptions in women younger than 18 years, which was greater in areas of greater deprivation and also in areas of higher Teenage Pregnancy Strategy investment.The steep deprivation gradient previously associated with conception in women younger than 18 years has been partly attenuated.Similarly, the association between conception before age 18 years and lower educational level remains signifi cant but has weakened over the period.Young  people increasingly learn about sexual matters mainly from school lessons, and the association between conception before age 18 years and receiving sex education from other sources remains strong.The prevalence of participation in education, work, and training in women with a child conceived before age 18 years, although low, increased between 1999-2001 and 2010-12, a trend not seen in other, same-aged women.
Reductions in rates of teenage pregnancy-ie, those occurring before age 20 years-have been seen in other high-income countries, 13 indicating a broader secular trend.Fewer comparative data exist specifi cally on pregnancies in women younger than 18 years, and hardly any include abortion data.However, data for the birth rate in women younger than 18 years in the European Union between 2004 and 2014 show a decline from 13    21 and highlight the importance of education in planning policies with this as their aim. 22his was refl ected in the decision by the English Government to locate the unit responsible for the Teenage Pregnancy Strategy jointly in two government ministries, the Department of Education and the Department of Health, and to make joint working between education and health agencies a key component of the strategy.
A strength of this study lies in the combination of individual-level and area-level data, and in our capacity to show independent associations between conception rates in women younger than 18 years and possibly infl uential variables.However, slight changes in question wording between the surveys might have infl uenced responses relating to pregnancy in Natsal-2 and Natsal-3.In particular, we should note that abortions were slightly under-reported in Natsal-3 compared with the offi cial UK fi gure in 2011. 23ompared with experimental approaches, observational studies have limitations in assessing the relative eff ect of policy-related intervention and secular trends on health outcomes.Specifi cally in this case, Teenage Pregnancy Strategy funding was determined by the pre-intervention conception rate.This creates challenges in disentangling the eff ect of Teenage Pregnancy Strategy funding from the eff ects of the pre-intervention rate on the subsequent conception rate, including the potential for regression to the mean.Our fi nding of an association between the decline in conception rates in women younger than 18 years and Teenage Pregnancy Strategy resourcing is suggestive that government-linked eff orts have contributed towards lowering conception rates in women younger than 18 years but should not be seen as conclusive.
Individual-level-data have also enabled us to show, where analyses of area-related data have not, 24 the progress that has been made in reducing the previously strong link between deprivation and early conception.The comparatively greater decline in conception rate in women younger than 18 years in the most deprived areas is worthy of note.Although the potential for decline might have been greater where baseline conception rates were higher, these were often areas in which complex and multiple social and health problems competed for public health eff orts and resources.
Progress has been made towards halting the cycle of poverty and income inequality long associated with early pregnancy, and in improving the life chances of young mothers.Despite this success, England's teenage pregnancy prevalence remains high compared with other high-income countries, 25 and there is more to be achieved.Some have suggested, because improvements in use of eff ective contraception seem to have contributed to the decrease in rates, that the policy emphasis should be placed where it has been successful in the past.Higher contraception discontinuation rates in young people 26 favour their use of long-acting reversible contraception and additional progress could be made by accelerating the rate at which they are used after onset of sexual activity.Despite the increase in use of these methods, they are used by barely one in six teenage women.The strong and sustained association between conception in women younger than 18 years and earlier sexual activity suggests that sizeable additional gains might also be made by helping young women to become sexually active at a time that is right for them. 27A third of all young women, and 60% of those for whom fi rst sex occurs before age 16 years, subsequently consider that to have been too early for them.Bringing actual timing of fi rst sexual encounter in line with preference for timing of fi rst sexual encounter could cut teenage pregnancy rates further, and an intensifi ed policy focus aimed at achieving this might be warranted.
Our data also underline the importance of continued eff orts to prevent teenage pregnancy.The decrease in conception rates in women younger than 18 years was more substantial in the later years of the Teenage Pregnancy Strategy and before then, there was scope for detractors to dismiss it as having failed to reach its goals. 28,29Our interim study of the assessment of the Teenage Pregnancy Strategy in 2006 showed modest falls in the fi rst 4 years of the strategy, albeit more marked in poorer areas of the country 30 in which government investment had been greatest.From 2007, however, pregnancy rates began to decrease more steeply and the downward trend continued after the strategy was mainstreamed in 2010.The acceleration is partly explained by the likelihood that strategy-related intervention took time to be implemented.It might also refl ect the renewed eff orts mounted in 2006 to mobilise additional resources and to persuade senior stakeholders to prioritise the issue in areas in which multiple socioeconomic problems were impeding progress, which is coincidental with a steeper decline in conception rates in women younger than 18 years in these areas in the later years of the Teenage Pregnancy Strategy.A further factor possibly contributing to the decrease in conception rates in women younger than 18 years is the synergy between strategy-related initiatives and other therapeutic and policy-related intervention-the availability of emergency contraception without prescription, for example, the recommendation by the National Institute for Health and Care Excellence that access to long-acting reversible contraception should be increased, and the rise in educational attainment in young people.Taken as a whole, the evidence underlines the importance of long-term, sustained, multifaceted prevention strategies to tackle the more intractable public health challenges.

Table 1 :Figure 2 :
Figure 2: Conception in women younger than 18 years by quartile between 1994 and 2013 by Index of Multiple Deprivation (A) and Local Implementation Grant investment for Teenage Pregnancy Strategy (B)