为什么美国少女怀孕率下降？禁欲、性活动和避孕使用的作用 Why is Teenage Pregnancy Declining? The Roles of Abstinence, Sexual Activity and Contraceptive Use http://www.agi-usa.org 概况（Introduction） 1990年代少女生育率持续下降，鉴于少女生孩子对年轻妇女及其孩子经常产生消极后 果，也是社会的一个负担，因此这是一个激动人心的发展（注1）。但是，少女生育不应该 是孤立的问题。每年四分之三以上的少女怀孕不是想要的，四分之一以上的少女怀孕结局 是堕胎（注2）；因此，帮助年轻妇女避免非意愿怀孕也是一项重要的公共政策目标。再一 次，这方面的消息是鼓舞人心的；1990年代少女怀孕率也下降了（注3）。 但是，美国每年大约有90万20岁以下的少女怀孕，并且美国少女怀孕率是发达世界 中最高的（注4）。确保少女怀孕率保持下降趋势是必要的，一个关键步骤是理解已经取得 进步背后的因素。 少女怀孕下降可以通过两个途径实现，性行为的变化和避孕使用的变化。一些观察家 已经声称，下降是更多人禁欲的结果（注5）。还有人相信，这是因为青少年中更多人禁欲 和更多人使用避孕，特别是安全套使用，但是没有对率的下降的特定因素进行量化（注6）。 广泛的社会因素是这两个途径的基础。对艾滋病的恐惧、性态度的变化以及可以得到 新避孕技术，可能影响性活动，并改变那些发生性交的青少年使用避孕方法的模式。强劲 的经济改善了年轻人就业机会，福利改革对公共援助接受者进行限制，这些可能影响上述 行为，因为更多的教育和就业机会与较低的少女怀孕率和生育率是相关的。另外，许多干 预行动鼓励年轻人推迟性活动，和在有性活动时有效使用避孕措施。虽然有必要更多地评 价和监督这些项目，但是一些项目已经显示出很好的效果（注7）。很显然，对于制定旨在 防止少女怀孕的政策和干预，确定并量化这些因素如何影响青少年性行为、避孕使用和怀 孕率，这是有用的。但是，这种探索已经超出了本文的范围。 但是，作为第一步，本报告给出了对最新资料的分析结果，记录了少女怀孕下降的幅 度，分析了禁欲、那些曾经性交过的青少年的性行为和避孕使用的变化对这些趋势的贡献。 分析是根据1988年和1995年“全国家庭增长调查”（National Survey of Family Growth， NSFG）的资料和最近关于少女怀孕率、生育率和流产率的资料。 趋势是什么？（What Are the Trends?） 自从1990年代，少女怀孕率、生育率和流产率显著下降；怀孕率和流产率下降到自1970 年代首次测量来的最低点，生育率接近1970年代中期和1980年代中期之间的常见水平。（注 8） 怀孕率和流产率下降地特别陡峭。1986年，每1000名15-19岁妇女大约有107个怀 孕，到了1990年，这个率上升11%，到每1000名15-19岁妇女有117个怀孕（图1）（注 9）。但是，随后6年中，这个率下降17%，到每1000名少女有97个怀孕，比1986年的 率低9%。 少女生育率跟随相似趋势，虽然最近下降不那么快。1986年，每1000名15-19岁妇 女有50个生育，这个率快速上升到1991年每1000名62个生育，上升了24%。随后5年 形势出现扭转，1996年，每1000名少女发生54个生育，这个率依然高于1986年，但是 比1991年的高峰下降12%。 少女流产率模式稍有不同，1980年代流产率变化很小，随后开始稳定下降。到1996 年，这个率是每1000名少女有29次流产，比10年前少女流产率下降31%。 下降是普遍的（Declines Are Widespread） 对不同分组青少年资料的分析揭示，无论年轻妇女的年龄、婚姻状况、宗族或民族， 几乎没有例外，怀孕率趋势全部跟随同样的模式。 在18-19岁的少女中，怀孕率从1980年代后期到1991年上升8%，到1996年，怀孕 率下降12%，低于高峰水平（图2）（注10）。生育率跟随相似模式，1991年和1996年之 间下降9%。这些年龄最大的少女中的流产率在1980年代基本徘徊在每1000名妇女有60 次流产，从1990年开始下降，到1996年下降到每1000名妇女45次流产，大约下降25%。 同样，对于15-17岁组，怀孕率1991年达到高峰，然后下降，到1996年下降16%（图 3）（注11）。这个期间，生育率下降13%，这些年轻妇女的流产率在1980年代后期开始下 降，1996年的流产率比10年前低36%。 甚至在14岁及其以下年龄的少女中，她们的怀孕率总是很低，怀孕变得比1990年代 更少见。1996年这个年龄组每1000名女性发生大约13个怀孕，比1980年代中期到1993 年每1000名大约17个怀孕的比率下降了。（注12） 很小比例的已婚少女（1996年4%的15-19岁少女已婚（注13））也出现这些率的大幅 下降。1990年和1996年期间，已婚的15-19岁少女怀孕率下降19%，从每1000名535个 怀孕下降到432个怀孕，生育率下降18%，从每1000名420个生育下降到344个生育。 这些变化反映出，一方面已婚少女不那么会怀孕了，另一方面非婚怀孕的少女不那么会在 孩子出生前结婚了。虽然已婚少女流产率一直很低，但是1988年和1996年期间下降了43%， 从每1000名31个流产下降到18个流产。（注14） 非婚少女中的趋势跟随一个稍微不同的模式（图4）（注15）。1990年和1996年期间 怀孕率下降14%，而生育率在1986年和1990年代早期稳定上升，然后呈平稳状况，随后 于1994年和1996年期间下降8%。1996年的生育率依然比10年前高33%。同时，非婚少 女中的流产率下降31%。因此，生育率上升反映出，一方面非婚怀孕少女在生育前会结婚 的可能性降低了，一方面她们通过流产终止怀孕的可能性也降低了。 非西班牙裔白人少女历史上比黑人少女或西班牙裔少女怀孕（生育、流产）率低，这 个趋势1990年代持续着。而且，这些率在这个期间下降了（图5）；怀孕率和流产率下降 地比生育率快（分别是24%、41%与12%）（注16）。 1990年代大部分时间，黑人少女的率最高。但是，所有这三个率在1990年和1996年 期间下降了大约20%（图6）。因为黑人少女生育率比白人少女下降地快，这两个组的差距 缩小了。 相反，西班牙裔少女的率在1990年代早期增加了（图7）（注18）。怀孕率和流产率在 1993年后轻微下降，但是生育率直到1994年保持稳定，随后下降5%。结果，西班牙裔少 女现在有着最高的生育率。 美国少女不是唯一怀孕率、生育率和流产率下降的年龄组。20岁出头的妇女的率虽然 比少女的率高，但是也跟随同样的模式。20-24岁妇女怀孕率和生育率在1980年代后期上 升后，在1990年和1996年期间分别下降了7%和5%。这个年龄组的流产率甚至下降的更 大（11%）。20-24岁非婚妇女生育率上升43%，从1986年每1000名49个怀孕上升到1996 年每1000名71个怀孕。（注19） 这些趋势的原因是什么？（What Accounts for These Trends?） 公众注意力大量集中在少女生育率令人欣喜的下降，但是这里记录的广泛趋势显示， 这不是这副画面的唯一部分。生育率下降反映出少女怀孕水平以及这些怀孕解决方式的变 化。如果怀孕率下降，或者如果怀孕妇女更多地利用流产避免非意愿生育，生育率就会下 降。什么是少女生育率下降的主要原因？ 1991年和1996年期间，以流产终止少女怀孕的比例从38%下降到35%（图8）（注20）。 因此，少女生育率下降归因于她们怀孕率的下降。因此，问题变成什么是怀孕率下降的原 因？ 所有少女怀孕率数学上是两个因素的产物：因为发生过性交而有潜在怀孕危险的少女 比例，以及因为发生性交而怀孕的少女比率。因此，分析这些因素随着时间的变化是理解 每个因素多大程度上对整个少女怀孕率贡献的关键。 为这种分析可以得到的信息在许多方面受到限制，包括只有1988年和1995年的资料 （而不是1990-1996年的资料，这个期间怀孕率下降最快），依据的少女样本太小，不能准 确发觉很小的变化（注21）。因此，这次分析探索了1988年和1995年期间总体少女怀孕 率变化的原因；但是，这些趋势可能在不同分组和不同时间段有所不同。 性行为和怀孕率的变化对下降发挥作用(Changes in Sexual Behavior and Pregnancy Rates Contributed to Declines) 1988年和1995年间，少女怀孕率下降10个百分点，从每1000名少女111个怀孕下 降到101个怀孕。根据“全国家庭增长调查”（NSFG），这个期间，发生过性行为的少女 比例下降2%，从52.6%降到51.3%（注22）。另外，这些年来有性经验少女怀孕的情况大 幅下降，其怀孕率从每1000名（有性经验少女）212个怀孕降到197个怀孕（图9）（注23）。 发生过性行为少女比例和有性经验少女中怀孕率的变化的相对作用可以通过计算假如 只有一个因素变化怀孕率会是怎样来估算。这种分析假定，1988年和1995年期间15-19 岁少女发生过性交的比例下降是真实的，即使统计学上并不显著。这种计算发现，大约四 分之一1988年和1995年期间少女怀孕率下降是因为更多人禁欲（也就是发生过性行为少 女比例下降），大约四分之三是因为有性经验少女怀孕率下降了。（参见下表）（注24） 计算禁欲和有性经验少女怀孕率变化所起作用（Calculating the Contributions of Changes in Abstinence and in the Pregnancy Rate of Sexually Experienced Teenagers） 用总体少女怀孕率除以有过性经验少女的比例得出15-19岁有过性经验少女怀孕率。 因此，1988年，当时怀孕率是每1000名15-19岁少女有111.4个怀孕，这个年龄组少女有 52.6%发生过性交，有过性经验的15-19岁少女怀孕率是每1000名有211.8个怀孕 （111.4/0.526=211.8）。1995年，总体少女怀孕率是每1000名有101.1个怀孕，51.3%15- 19 岁少女发生过性交，因此，15-19岁有性经验少女怀孕率是每1000名197.1个怀孕 （101.1/0.513=197.1）。 如果有过性经验少女比例下降了（从1988年52.6%降到1995年51.3%），但是15-19 岁有过性经验少女怀孕率保持不变（每1000名211.8个怀孕），那么1995年总体少女怀孕 率（上述两个因素的产物）就会是每1000名108.7个怀孕（0.513x211.8=108.7）。这将表 示 怀孕率从1988年每1000名111.4个怀孕减少2.7个怀孕。因此，每1000名少女实际减少 的10.3个怀孕中大约25%是因为有性经验少女比例减少。 相反，如果15-19岁有过性行为少女比例保持在1988年52.6%的水平，而有过性经验 少女怀孕率从1988年每1000名211.8个怀孕下降到1995年197.1个怀孕），那么1995年 总体怀孕率就会是每1000名少女103.7个怀孕（0.526x197.1=103.7）。这次每1000名减少 7.7个怀孕是因为，有性经验少女怀孕率下降大约是整个观察到的每1000名15-19岁少女 怀孕率下降的75%。 注释和资料来源参见注24。 为什么有性经验少女怀孕率下降？（Why Has the Pregnancy Rate Among Sexually Experienced Teenagers Declined?） 进一步探索的一个重要问题是有性经验的少女如何更加成功地避孕。她们是否通过减 少性交的频数而降低了暴露于怀孕的机会？或者，她们是否通过增加使用避孕方法或更加 有效的避孕措施而降低发生性行为时怀孕的危险性（注25）？上述因素的任何结合可能产 生作用，这里的焦点是什么样的行为改变对怀孕率下降发挥作用，而不是什么导致了行为 改变。 平均性交频数改变（Average Frequency of Sexual Intercourse Has Changed） 虽然改变青少年性行为的注意力主要集中在增加禁欲，而其它措施则给少女带来怀孕 危险。但是，有性经验少女平均接触性交在最近几年几乎没有变化。例如，大约79%参与 1995年全国家庭发展调查的有性经验少女报告在过去三个月中发生性行为，而1988年是 81%（注26） 更加重要的是，1995年比1988年有更大比例的有性经验少女报告在过去一年没有发 生性行为（分别是9%与5%），有更小比例的（有性经验少女）过去一年在一些月份而不 是所有月份发生性交（分别是49%与60%），有更大比例的（有性经验少女）说，她们一 年中每个月发生性交（分别是43%与35%）（图10）（注27）。这些不平衡变化的结果是， 有性经验少女过去一年中发生性交的平均月数没有变化--8.6个月（注28）。 类似的是， 1988年和1995年间有性经验少女发生性交的频率几乎没有变化。在这两 年中，大约一半在过去三个月中发生性行为的少女报告至少每周发生性交，虽然报告每周 性交4次或以上的比例从4%增加到10%。（注29） 利用上述一些性行为计量方法，而不是曾经发生过性行为少女比例，得出了改变性行 为对1988年和1995年间怀孕率下降作用的不同分配，范围从总体下降的0%到65%。上 述其它计量方法最相关的是用少女平均发生性交月数校正过去一年发生性行为的15-19岁 少女比例。用这个计量方法计算和用禁欲水平变化计算显示出同样的影响，大约四分之一 总体怀孕率下降是因为性接触减少，四分之三是因为发生过性行为少女怀孕率下降（注 30）。 避孕使用的趋势是复杂的（Trends in Contraceptive Use Are Mixed） “全国家庭增长调查”资料提供了三个点上少女避孕使用的快照：她们第一次发生性 交、调查进行的那个月和她们最后一次发生性交。最大的变化是有性经验少女报告在第一 次性交时使用一种（避孕）方法的比例增加了。1995年大约75%给出这个回答，相比之下， 1988年是65%（图11）（注31）；这个改善最主要是因为第一次性交时安全套使用从48% 上升到63%。 但是，第一次性交时（避孕）方法使用反映了一次经验，而对处于非意愿怀孕危险（比 如发生性行为、有生育能力、没有怀孕、产后或试图怀孕）的少女中随后（避孕方法）使 用的测量显示不那么令人鼓舞的迹象。比如，1988年和1995年，处于危险的少女说自己 在被采访这个月使用避孕的比例仅仅增加3%，从78%生到80%。（注32） 和上述增加相反，过去三个月发生性交并在最后一次性行为时使用避孕方法的少女比 例从85%下降到83%。下降尤其是非激素、性交有关的避孕方法（提前撤出、杀菌剂和安 全期），而不是安全套。（注33） 少女使用更加有效的方法（Teenagers Are Using More Effective Methods） 少女避孕使用者越来越多地采用长效激素方法（可注射和植入），这些方法在所有可逆 方法中失败率最低。这些1990年代早期才出世的方法占1995年避孕使用的13%（图12）： 大约10%的少女使用者依靠可注射的，3%依靠可植入的。安全套使用稍有增加，而对口服 避孕药的依赖大幅下降。（注34） 主要因为转向长效方法，少女中总体避孕效果在1988年和1995年期间改善了，或者， 换句话说，少女避孕使用者变得不太容易怀孕了。在1988年避孕使用者的方法模式下，据 估计16%在开始使用一年内怀孕；到1995年，比例下降到15%，大约下降9%。（注35） 政策意义（Policy Implications） 最近美国少女怀孕率下降是一个非常让人高兴的发展，下降的原因需要大量的思考。 通过记录少女生育行为变化的成分，本报告朝着进一步改善迈出了重要的第一步。有资料 表明，1991年以来少女生育率下降，因为15-19岁有性经验少女怀孕率下降了，而不是流 产率上升了。 1970年代和1980年代早期有性经验少女避孕的改善被发生性行为少女比例稳定上升 平衡掉了。从而导致这个期间总体怀孕率上升或稳定。（注35） 这次分析表明模式的重要变化：现有资料表明，1988年和1995年期间两个因素在同 一个方向发挥作用，共同对少女总体怀孕率下降做出贡献。结论是尝试性的，因为性行为 变化太少，不能认为统计学上有意义，即使在相对较大的“全国家庭增长调查”全国样本， 以及因为两次调查时间间隔很长，但是这些资料显示，从1970年代和1980年代早期的模 式中出现变化。 好象是，更多少女禁欲解释大约四分之一1988年和1995年期间美国少女怀孕率的下 降。 没有禁欲的少女性活动水平变化可能几乎没有对怀孕率发生任何作用，如果有任何作 用的话，因为发生过性交少女性活动的一些测量上的变化被相反方向的变化抵消了。 避孕使用水平趋势是复杂的，但是少女避孕使用转向新的可以得到的长效激素方法， 主要是可注射避孕药，这是有性经验少女避孕效果提高上的一个显著变化。 这些研究结果表明，性活动减少和使用更有效避孕方法在最近少女怀孕率和生育率下 降上发挥作用。无疑，这两个行为变化受到广泛社会变化的影响，这些变化反映在政策、 项目、态度和价值观上。我们还没有理解这些变化及其相互联系的许多方面。 即使这样，这些研究结果提示，维持少女怀孕水平下降的最好策略是多种方法的结合。 项目和政策应该旨在鼓励青少年，特别是那些低龄青少年，推迟性交，以及支持有性经验 的青少年不要有进一步的性活动。同时，必须认识到，大多数年轻人在少年期有性活动， 性教育和知识必须也为她们/他们准备足够的预防怀孕和性传播感染知识和能力，如果并当 她们/他们发生性行为。必须提供服务，帮助她们/他们行为上负责任，确保她们/他们使用 避孕措施，帮助她们/他们改善避孕效果。这意味着提供足够的关于性行为及其后果的教育 和知识，以及提供匿名的、可以负担的和容易得到避孕服务和供应，支持年轻人更容易接 受并容易有效使用的新避孕法研究和开发。 虽然青少年必须面对和管理大量压力和需求，但是她们/他们必须对自己行为负责。同 时，这也同样是决策者、教育者、父母和社会的责任，应该帮助她们/他们作好准备做这些 事情，让环境尽可能更有助于她们/他们成功地做这些事情。 References and Notes 1. Ventura SJ, Matthews MS and Curtin SC, Declines in teenage birth rates, 1991- 1998: update of national and state trends, National Vital Statistics Reports, 1999, Vol. 47, No. 26; The Alan Guttmacher Institute (AGI), Sex and America's Teenagers, New York: AGI, 1994; and Maynard RA, Kids Having Kids, Washington, DC: Urban Institute Press, 1997. 2. Note: A pregnancy is defined as unintended if the woman had an induced abortion or if she gave birth at a time she had wanted not to have a baby. Source: Henshaw SK, Unintended pregnancy in the United States, Family Planning Perspectives, 1998, 30(1):24-29 & 46. 3. Henshaw SK, U.S. teenage pregnancy statistics, with comparative statistics for women aged 20- 24, New York: AGI, 1999. 4. Ibid.; and Singh S and Darroch JE, Adolescent pregnancy and childbearing: levels and trends in industrialized countries, Family Planning Perspectives, 2000, 32 (forthcoming). 5. Jones JM et al., The declines in adolescent pregnancy, birth and abortion rates in the 1990s: what factors are responsible? Fanwood, NJ: Consortium of State Physicians Resource Councils, 1999. 6. Donovan P, Falling teen pregnancy, birthrates: What's behind the declines? The Guttmacher Report on Public Policy, 1998, 1(5):6-9; and Ventura SJ et al., Declines in teenage birth rates, 1991-97: national and state patterns, National Vital Statistics Reports, 1998, Vol. 47, No. 12. 7. Note: The federal-state abstinence-only funding that was part of welfare reform legislation began in FY 1998, after the time period for which pregnancy data are available. (Sources: PL 104- 193, Section 912; and Daley D and Wong VC, Between the Lines: States' Implementation of the Federal Government's Section 510(b) Abstinence Education Program in Fiscal Year 1998, New York: Sexuality Information and Education Council of the United States, 1999.) Sources: Card JJ et al., The Program Archive on Sexuality, Health & Adolescence: promising 'prevention programs in a box,' Family Planning Perspectives, 1996, 28(5):210-220; Kirby D, No Easy Answers: Research Findings on Programs to Reduce Teen Pregnancy, Washington, DC: The National Campaign to Prevent Teen Pregnancy, 1997; Moore KA et al., Adolescent Pregnancy Programs: Interventions and Evaluations, Washington, DC: Child Trends, 1995; AGI, 1994, op. cit. (see reference 1); and Landry DL, Kaeser L and Richards CL, Abstinence promotion and the provision of information about contraception in public school district sexuality education policies, Family Planning Perspectives, 1999, 31(6):280-286. 8. Henshaw SK, 1999, op. cit. (see reference 3); Ventura SJ, Matthews MS and Curtin SC, 1999, op. cit. (see reference 1). 9. Notes: All rates refer to age at time of outcome. Pregnancies include miscarriages, estimated as 20% of births plus 10% of abortions. (Source: Leridon H, Human Fertility: The Basic Components, Chicago: University of Chicago Press, 1977, Table 4.20.) Source: AGI, Teenage pregnancy: overall trends and state-by-state information, New York: AGI, 1999. 10. Note: See reference 9. Source: Henshaw SK, 1999, op. cit. (see reference 3). 11. Note: See reference 9. Source: Henshaw SK, 1999, op. cit. (see reference 3). 12. Notes: See reference 9. The numerators for rates among women 14 and younger include pregnancies, births and abortions among all those younger than 15 at outcome; the denominators are females aged 14. Since some pregnancies, births and abortions occurred among women younger than 14, these rates slightly overstate the actual rates among 14-year-olds. For annual rates, see Table 1. Source: Henshaw SK, 1999, op. cit. (see reference 3). 13. Saluter AF and Lugaila TA, Marital status and living arrangements: March 1996, Current Population Reports, 1998, Series P-20, No. 496, Table 1. 14. Notes: See reference 9. All rates refer to marital status at time of outcome. The rates shown here understate conception rates among unmarried teenagers (those who were never-married, divorced or widowed) and overstate rates among married teenagers (including those who were separated from their husband). Of first births to women who were married at delivery, only 52% in 1985-1989 and an estimated 45% in 1990-1994 resulted from conceptions occurring within marriage; of first births resulting from out-of-wedlock conceptions, 19% in 1985-1989 and 16% in 1990-1994 were to women who married before delivery. (Sources: Bachu A, Fertility of American women: June 1990, Current Population Reports, 1995, Series P-20, No. 454, Table E; and Bachu A, Timing of first births: 1931-34 to 1990-94, Working Paper, Washington, DC: Population Division, U.S. Bureau of the Census, 1998, No. 25, Figures 9 and 10.) The apparent dip in marital birthrates and pregnancy rates in 1994 may result from changes in the reporting and estimation of marital status for births. (Sources: Ventura SJ et al., Advance report of final natality statistics, 1994, Monthly Vital Statistics Report, 1996, Vol. 44, No. 11 (Supplement), pp. 82-83; and Ventura SJ et al., Advance report of final natality statistics, 1995, Monthly Vital Statistics Report, 1997, Vol. 45, No. 11 (Supplement), pp. 7-8.) For annual rates, see Table 1. Sources: Nonmarital and marital births and birthrates for 1986-1993--Ventura SJ et al., The demography of out-of-wedlock childbearing, in: Department of Health and Human Services (DHHS), Report to Congress on Out-of-Wedlock Childbearing, Hyattsville, MD: DHHS, 1995, Table I-2, p. 88, and Table III-7, p. 118; and AGI, 1999, op. cit. (see reference 9). Nonmarital birth and birthrates and marital births for 1994-1996--Ventura SJ et al., Advance report of final natality statistics, 1994, Monthly Vital Statistics Report, 1996, Vol. 44, No. 11 (Supplement), Tables 2, 3 and 14; Ventura SJ et al., Advance report of final natality statistics, 1995, Monthly Vital Statistics Report, 1997, Vol. 45, No. 11 (Supplement), Tables 2, 3 and 14; and Ventura SJ et al., Advance report of final natality statistics, 1996, Monthly Vital Statistics Report, 1998, Vol. 46, No. 11 (Supplement), Tables 2, 3 and 17. Marital birthrates for 1994-1996--Ventura SJ, National Center for Health Statistics, personal communication, June 1999. Marital-specific abortion rates for 1986- 1996--calculated for each year by multiplying the marital-specific birthrates by marital-specific ratios of numbers of abortions to numbers of births. The numbers of abortions by marital status were estimated by applying the distributions of abortions to women 15-19 by marital status, interpolated for each year from distributions in 1987 and 1994 (unpublished tabulations from the 1987 and 1994-1995 AGI Abortion Patient Surveys), to the annual number of abortions among all women aged 15-19 (Henshaw SK, 1999, op. cit.--see reference 3). 15. See notes and sources in reference 14. 16. Notes: See reference 9. Abortion and pregnancy data by Hispanic ethnicity are available only for 1990 and later. Rates for non-Hispanic women of races other than white and black are not shown because of small numbers, but these women are included in other figures presented here. Because race and ethnicity for women giving birth and having abortions are sometimes reported by someone other than the woman, they may not always be consistent with the race and ethnicity that the woman would have reported for herself. Sources: Birthrates--Ventura SJ et al., Declines in teenage birth rates, 1991-97: national and state patterns, National Vital Statistics Reports, 1998, Vol. 47, No. 12, Table 2. Abortion rates among black teenagers--AGI, 1999, op. cit. (see reference 9). Rates for non-Hispanic white and Hispanic teenagers--Henshaw SK, AGI, unpublished tabulations, 1999. 17. Ibid. 18. Ibid. 19. Notes: See reference 9. For annual rates for women aged 20-24, see Table 1. Sources: Henshaw SK, 1999, op. cit. (see reference 3); and Ventura SJ et al., Advance report of final natality statistics, 1996, Monthly Vital Statistics Report, 1998, Vol. 46, No. 11 (Supplement), Table 18. 20. AGI, 1999, op. cit. (see reference 9). Miscarriages are not included in the denominator. 21. National data on sexual behavior and contraceptive use among women aged 15-19 are available from the 1988 and 1995 NSFGs. A 1990 follow-up to the 1988 survey more closely coincides with the peak teenage pregnancy rates; however, it differed in important ways from the 1988 and 1995 surveys, and therefore is an inappropriate source of data for examining time trends. Whereas the 1988 and 1995 surveys used comprehensive personal household interviews, the follow-up survey was conducted through short telephone interviews; this difference may have affected teenagers' reporting of sensitive topics, such as sexual behavior and contraceptive use. Additionally, low response rates to the 1990 survey limit the generalizability of the data. The response rate for the 1988 NSFG was 79%, allowing for nonresponse to both the National Health Interview Survey (NHIS), from which the NSFG sample was drawn, and the NSFG itself. (Source: Mosher WD, Contraceptive practice in the United States, 1982-1988, Family Planning Perspectives, 1990, 23(5):190-205.) The 1990 telephone survey reinterviewed only 68% of women aged 17-44 who had responded to the 1988 NSFG, or about 55% of the original sample. A subsample of women aged 15-17 in 1990 was added from NHIS respondent households; 53% of these women participated in the 1990 interviews. (Source: Peterson LS, Contraceptive use in the United States: 1982-1990, Advance Data from Vital and Health Statistics, 1995, No. 260.) Respondents to the 1995 NSFG represented 81% of women aged 15-17 and 77% of women aged 18-44 selected from the 1993 NHIS, which had a response rate of about 95%, for overall response rates of roughly 77% and 73%, respectively. (Sources: Kelly JE et al., Plan and operation of the 1995 National Survey of Family Growth, Vital and Health Statistics, 1997, Series 1, No. 35; and Mosher WD, Design and operation of the 1995 National Survey of Family Growth, Family Planning Perspectives, 1998, 30(1):43-46.) 22. Notes: Sexually experienced women include all those aged 15-19 at the time of their NSFG interview who had ever had intercourse (voluntary or involuntary) since menarche. The change between 52.6% in 1988 and 51.3% in 1995 is not statistically significant; the 95% confidence intervals are 47.8-57.4% for the 1988 estimate and 48.4-54.2% for 1995. (Source: Tabulations from the 1988 and 1995 NSFGs.) Generally, the proportion of women who had ever had intercourse since menarche did not change significantly from 1988 to 1995 for subgroups of women by age or race and ethnicity; the exception was that Hispanic women aged 15-17 were more likely to have had intercourse in 1995 than in 1988. The estimates of sexual exposure are quite sensitive to variable definitions. The proportion of women aged 15-19 in the 1988 NSFG who had ever had intercourse since menarche is 52.9% when age is based on the respondents' age at the actual date used to select the sample (March 15, 1988), but 52.6% when it is based on the respondent's age at the actual date of the interview. In the 1995 NSFG, the proportion of 15-19-year-olds who ever had intercourse (voluntary or involuntary) since menarche is 50.4% when age is based on the sample selection date (April 1, 1995), but 51.3% when based on the date of interview. The 1988 NSFG did not distinguish voluntary from involuntary intercourse; in the 1995 survey, 51.0% of women aged 15-19 at the date of their interview had had only voluntary intercourse since menarche. Including young women who had had intercourse only prior to menarche would increase the proportions who had ever had intercourse (based on age at interview) from 52.6% to 52.9% in 1988 and from 51.3% to 51.7% in 1995. Sources: Singh S and Darroch JE, Trends in sexual activity among adolescent American women: 1982-1995, Family Planning Perspectives, 1999, 31(5):212-219; and tabulations from the 1995 NSFG. 23. Note: See reference 9. Sources: AGI, 1999, op. cit. (see reference 9); Singh S and Darroch JE, 1999, op. cit. (see reference 22); and tabulations from the 1995 NSFG. 24. Notes: In addition to effects of changes in the proportion of women aged 15- 19 who ever had intercourse (25%) and in the pregnancy rate of sexually experienced young women (75%), the interaction of these two factors accounted for -2% of the change in the overall pregnancy rate. These calculations are rough estimates because there is a margin of error around the measures of sexual activity (see notes in reference 22). The proportion of 15-19-year-olds who had ever had intercourse would have to have decreased from 52.6% in 1988 to 47.7% in 1995 for increased abstinence to have accounted for the entire pregnancy rate decline. This is slightly above the 1982 level and outside the 95% confidence interval for the 1995 proportion who had ever had sex; that is, a change this large would have been identified in the NSFGs. For increased abstinence to have accounted for half of the pregnancy rate drop, the proportion sexually experienced would have to have decreased to 50.2% in 1995; since that proportion is within the 95% confidence interval for that year, a change of this magnitude cannot be ruled out, given the margin of error around the NSFG estimates. Sources: Pregnancy rates--AGI, 1999, op. cit. (see reference 9). Proportions of young women who ever had sexual intercourse (voluntary or involuntary) since menarche--Singh S and Darroch JE, 1999, op. cit. (see reference 22); and tabulations from the 1995 NSFG. 25. Notes: Effects of other factors could also influence the pregnancy rates of sexually experienced teenagers, but are unlikely to have done so or cannot be measured with available data. For example, the biological fecundity of young women or their partners could have decreased. However, almost all women aged 15-19 have experienced menarche, and the age of menarche appears to have continued to drop over the period in question. (Sources: Forrest JD, Timing of reproductive life stages, Obstetrics and Gynecology, 1993, 82(1):105-111; and Abma J et al., Fertility, family planning, and women's health: new data from the 1995 National Survey of Family Growth, Vital and Health Statistics, 1997, Series 23, No. 19, Table 18.) Similarly, while a number of reports have suggested that sperm counts in the United States are declining, further analysis indicates that no significant change has occurred during the last 50 years. (Source: Saidi JA et al., Declining sperm counts in the United States? A critical review, Journal of Urology, 1999, 161(2): 460-462.) The proportion of sexually experienced women aged 15-19 who have had two or more sexual partners increased from 58% in 1988 to 63% in 1995, according to data from the comparable, interview-administered portion of the survey. (Sources: Kost K and Forrest JD, American women's sexual behavior and exposure to risk of sexually transmitted diseases, Family Planning Perspectives, 1992, 24(6):244-254; and Abma J et al., Fertility, family planning, and women's health: New data from the 1995 National Survey of Family Growth, Vital and Health Statistics, 1997, Series 23, No. 19, Table 30.) The number of sexual partners is less strongly related to the risk of pregnancy, however, than is the frequency of intercourse. (Source: Cates W Jr., Sexually transmitted diseases and family planning: strange or natural bedfellows? Sexually Transmitted Diseases, 1993, 20(3): 174-178.) 26. Note: The 1995 proportion of women 15-19 who had sex in the three months prior to interview was calculated both from direct questions and from information in the partnerships and contraceptive history sections of the interview. These calculations exclude women whose only sexual experience after menarche was involuntary. Source: Tabulations from the 1988 and 1995 NSFGs. 27. Tabulations from the 1988 and 1995 NSFGs. 28. Ibid.; and Singh S and Darroch JE, 1999, op. cit. (see reference 22). 29. Note: Among women aged 15-19 at the time of their NSFG interview who had had sex in the three months prior to interview and who provided information on frequency, 3.7% in 1988 had had intercourse four or more times a week during that period; 29.0%, 2-3 times a week; 18.0%, once a week; 29.1%, 2-3 times a month; and 20.2% once a month or less. In 1995, the proportions were 9.5% having had sex four or more times a week; 24.1%, 2-3 times a week; 16.6%, once a week; 26.9%, 2-3 times a month; and 22.6%, once a month or less. Source: Tabulations from the 1988 and 1995 NSFG. 30. The calculations for apportioning the change in the overall pregnancy rate to changes in sexual behavior and in the pregnancy rate of those having sex use the same methodology as that described above (see box, page 8). The key points for each mea-sure of sexual behavior are outlined below. Had sex in the three months prior to interview: The proportion of women aged 15- 19 at the NSFG survey date who had sex in the prior three months was 42.6% in 1988 and 40.3% in 1995, yielding pregnancy rates per 1,000 women aged 15-19 who had sex in the three months prior to interview of 261.5 in 1988 (111.4/0.426) and 250.9 in 1995 (101.1/0.403). If only the proportion who had sex in the three months prior to interview had changed, the 1995 overall pregnancy rate would have been 105.4--6.0 points lower than the actual rate. Thus, using this measure implies that the change in sexual behavior was responsible for 58% (6.0/10.3) of the decrease in the pregnancy rate, while 44% was due to a lower pregnancy rate among those having intercourse (4.3/10.3). Interaction between the changes in this sexual exposure measure and pregnancy rate accounted for -2% of the overall pregnancy rate change. Had sex in all 12 months of the past year: Some 35.4% of women aged 15-19 at interview date in 1988 had had sex in all 12 months of the past year, equal to 18.6% of all women aged 15-19 (52.6x0.354). In 1995, the proportion of sexually experienced women aged 15-19 who had sex throughout the past year was greater than in 1988 (42.9%), as was the proportion of all women aged 15-19 (51.3%x0.429=22.0%). Since the proportion of women aged 15-19 who had sex throughout the prior year increased from 1988 to 1995, the contribution of the change in this measure of sexual behavior was in the direction of increasing, rather than decreasing, the overall pregnancy rate. Thus, using this measure, declining sexual activity played no part in the overall pregnancy rate decrease, while the decreasing rate of pregnancy among those having sex using this measure accounted for all of the decrease. Had sex in any month of the past year: Some 49.9% of women aged 15-19 at the date of their NSFG interview in 1988 had had sex during the prior year, as had 46.9% in the 1995 NSFG. Pregnancy rates per 1,000 women aged 15-19 who had sex in the prior year were 223.2 in 1988 (111.4/0.499) and 215.6 in 1995 (101.1/0.469). If only the sexual exposure measure had changed between 1988 and 1995, the overall pregnancy rate would have been 104.7 (0.469x223.2), or 6.7 points lower than the 1988 rate. Thus, on the basis of this measure, 65% of the actual overall pregnancy rate decrease (6.7/10.3) would have been due to a decrease in the proportion of women aged 15-19 who had sex in the 12 months before 1988 and 1995, and 37%, to a decrease in their pregnancy rate (3.8/10.3). The interaction would have accounted for -2% of the overall pregnancy rate decline. Exposure-adjusted measure of sex in the past year: The proportion of women aged 15-19 who ever had sex was 52.6% in 1988 and 51.3% in 1995. In each year, those who had ever had sex reported having had intercourse for an average of 71.7% of the prior year (an average of 8.6 of the prior 12 months). Multiplying the proportions sexually experienced by this average annual exposure measure yields adjusted proportions exposed of 37.7% in 1988 (52.6%x0.717) and 36.8% in 1995 (51.3%x0.717). Another way of expressing these proportions is that they are the average number of years of sexual exposure among all women aged 15-19--i.e., an average of 377 years of sexual activity per 1,000 women aged 15-19 in 1988 and an average of 368 years per 1,000 in 1995. The pregnancy rate using this measure--i.e., the number of pregnancies in a year divided by the number of woman-years of sexual activity per 1,000 women aged 15-19--was 295.5 in 1988 and 274.7 in 1995. If only sexual exposure had changed between 1988 and 1995, the overall pregnancy rate would have been 108.7 (0.368x295.5), a level 2.7 points below the actual 1995 overall pregnancy rate. Thus, 26% of the 10.3-point decline in the overall pregnancy rate (2.7/10.3) would have been due to decreased sexual exposure and 75% to the lower pregnancy rate among those having sex. The interaction effect would have been -2% of the overall decline. 31. Notes: The proportion in 1995 who had used contraceptives at first intercourse refers only to use at first voluntary intercourse among those who had had voluntary intercourse after menarche; those whose first intercourse after menarche was involuntary were not asked about contraceptive use on that occasion. The proportion of all sexually experienced teenagers who used contraceptives at first intercourse would have changed little if none or all those who only had had involuntary sex after menarche had used contraceptives (74.9% and 75.4%, respectively, compared with 75.3% of those who ever had voluntary intercourse).The increase between 1988 and 1995 in use of any method at first intercourse is statistically significant at p<.001. Source: Tabulations from the 1988 and 1995 NSFGs. 32. Note: The change in the level of current use among adolescents at risk of unintended pregnancy is not statistically significant. Source: Tabulations from the 1988 and 1995 NSFGs. 33. Note: The proportion who had used a method at last intercourse is calculated for women aged 15-19 at the interview date who had had sex in the last three months and were not pregnant, postpartum, trying to become pregnant or infertile at the time of the interview. The decrease in the proportion who had used a method at last intercourse is not statistically significant. In the 1995 NSFG, respondents who used no method in any of the three prior months or who used only a long-acting method (implant, injectable, IUD or contraceptive sterilization) were not asked about use at last intercourse. We have assumed that the women relying on a long- acting contraceptive used a method at last intercourse. In addition, 4% of current pill users in 1988 and 14% in 1995 reported using no method at last intercourse. Analysis of the 1995 survey showed most teenagers who used the pill in the survey month had, in fact, used it for all three recent months. (Source: Bankole A, tabulations from the 1988 and 1995 NSFGs.) Since a number of questions about consistency of pill use were added to the 1995 NSFG prior to the question about use at last intercourse and since oral contraceptives, like long-acting methods, require no action at the time of intercourse, we assumed that current pill users also had used their method at last sex, even if they said that they had used no method. The assumption that pill users who reported no method when asked about use at last intercourse were oral contraceptive users increased the 1995 proportion of women 15-19 at risk of unintended pregnancy using any method from 76.0% to 82.7%. Source: Tabulations from the 1988 and 1995 NSFGs. 34. Note: The change between 1988 and 1995 in the proportion of current contraceptive users relying on a long-acting method (implant, injectable or IUD) is statistically significant at p<.001. Source: Tabulations from the 1988 and 1995 NSFGs. 35. Note: Average pregnancy rates among contraceptive users were calculated by multiplying the 1988 and 1995 distributions of method users (tabulated from the 1988 and 1995 NSFGs) by the 1991-1994 first-year failure rates, according to method, poverty status and union status. Source: Fu H et al., Contraceptive failure rates: new estimates from the 1995 National Survey of Family Growth, . 36. AGI, 1994, op. cit. (see reference 1), Figure 20, page 41.