Are Births More Likely to be Intended Following Use of Long-Acting Reversible Contraceptives? An Analysis of U.S. Births in 2003–2015

Abstract

A major shift in the U.S. contraceptive method mix has been the recent growth in the use of long-acting reversible contraceptives (LARCs)—intrauterine devices and implants. Quantitative research into LARC’s impacts until now has focused on evaluating their efficacy in reducing unintended pregnancies. The next question, of whether births after discontinuing LARC use are then more likely to result from an intended pregnancy, has received almost no attention. We analyzed data from 2984 women who reported a live birth in the 3–4 years prior to survey interview for the 2006–2015 cycles of the National Survey of Family Growth. We compared the proportion of births intended by last contraceptive method used. To capture contraceptive failure versus stopping contraceptive use to become pregnant, we estimated logistic regressions alternately not controlling for, and controlling for, use of contraception in the month of conception. We found that four in five births following LARC use were reported to result from an intended pregnancy, compared to only three in five births following use of a moderately-effective or less-effective method. After controlling for use of contraception in the month of conception and for socio-demographic characteristics, women whose last-used method was a LARC had twice the odds of reporting that the pregnancy was intended relative to women whose last-used method was either a moderately-effective method or a less-effective method. We conclude that U.S. women’s LARC use has the potential to increase the fraction of subsequent births from intended pregnancies, and in doing so promote their reproductive autonomy.

Appendix

Detail on the Estimation of the Calendar Month of the Start of the Pregnancy and of the Last Contraceptive Method Used before the Pregnancy By Means of Alternative Survey Information for the 2015–2017 NSFG cycles.

Starting with the 2015–2017 NSFG, “century month date values for key life events have been suppressed from the public-use files to prevent potential linkage or use with external data sources to identify survey respondents.” (U.S. Department of Health & Human Services, 2018). These key life events include all pregnancies, meaning that the timing information for pregnancies in the 2015–2017 NSFG was limited to the calendar year of the start and end of the pregnancy—rather than the calendar month as in previous NSFG waves. To be able to include the latest NSFG data for the sensitivity analysis, the calendar month of the start of the pregnancy was estimated by relying on other information as reported by the respondent.

First, for each woman who reported the start of a pregnancy that resulted in live birth during a certain year covered by the contraceptive calendar, we considered whether there was a 9 + month period of non-use of contraception in the contraceptive calendar that started during that calendar year. If no 9 + month period of non-use of contraception was observed, a 6 + month period of non-use of contraception was considered instead. If we identified such a period of non-use of contraception, we took the first month of that period as being the month that the pregnancy started, with the only restriction being that this first month needed to be within the reported calendar year of the start of the pregnancy. This procedure has two potential issues:

• We might have taken a month before the actual month of the start of the pregnancy, if women did not use contraception for several months before the start of the pregnancy. However, this bias will not affect our ability to determine the last method used before the pregnancy started and is limited by the restriction that the estimated starting month needed to be within the reported calendar year of the start of the pregnancy.

• We might have taken a month after the actual month of the start of the pregnancy, if a woman continued use of contraception until after the pregnancy started (e.g., because she didn’t immediately know that she was pregnant). To examine how this bias may have affected our estimates, we created an alternative ‘starting month of the pregnancy’ variable by subtracting 3 months from the original value when determining the last contraceptive method used before the start of pregnancy. Sensitivity analysis using this alternative ‘starting month of pregnancy’ variable did not lead to substantively different conclusions (results available upon request).

A number of women did not have a 9 + month period or a 6 + month period of non-use of contraception that started in the reported calendar year of the start of the pregnancy. Most of these women appeared to continue to rely on methods such as condoms and withdrawal during the reported calendar year of the start and end of the pregnancy. Women for whom the starting month of the pregnancy could not be estimated, but who reported using only less-effective methods (potentially in combination with months with no contraceptive use) during the 12 months of the calendar year during which the pregnancy started, were included in the category ‘less-effective method’ of the variable last contraceptive method used before the pregnancy.

Based on these procedures, we were able to estimate the last contraceptive method used before the pregnancy for all but 78 out of 673 women who reported the start of a pregnancy resulting in live birth following use of a reversible contraceptive method during the contraceptive calendar in the 2015–2017 NSFG cycle.

See Tables 5

Table 5 Characteristics of mothers by whether they reported contraceptive use before the pregnancy resulting in live birth, U.S., 2003 to 2015

and 6

Table 6 Odds ratios (and 95% confidence intervals) of birth intendedness (reference outcome = unintended) in three sensitivity analyses, U.S., 2003 to 2015 or 2017

.