Critique of the Contraceptive CHOICE study

[fnr]

There’s a recent publication from the researchers conducting the “Contraceptive CHOICE” study, Provision of No-Cost, Long-Acting Contraception and Teenage Pregnancy. The story’s free on the New England Journal of Medicine’s web site.

Here is the abstract:

BACKGROUND
The rate of teenage pregnancy in the United States is higher than in other developed nations. Teenage births result in substantial costs, including public assistance, health
care costs, and income losses due to lower educational attainment and reduced earning potential.

METHODS
The Contraceptive CHOICE Project was a large prospective cohort study designed to promote the use of long-acting, reversible contraceptive (LARC) methods to reduce unintended pregnancy in the St. Louis region. Participants were educated about reversible contraception, with an emphasis on the benefits of LARC methods, were provided with their choice of reversible contraception at no cost, and were followed for 2 to 3 years. We analyzed pregnancy, birth, and induced-abortion rates among teenage girls and women 15 to 19 years of age in this cohort and compared them with those observed nationally among U.S. teens in the same age group.

RESULTS
Of the 1404 teenage girls and women enrolled in CHOICE, 72% chose an intrauterine device or implant (LARC methods); the remaining 28% chose another method. During the 2008–2013 period, the mean annual rates of pregnancy, birth, and abortion among CHOICE participants were 34.0, 19.4, and 9.7 per 1000 teens, respectively. In comparison, rates of pregnancy, birth, and abortion among sexually experienced U.S. teens in 2008 were 158.5, 94.0, and 41.5 per 1000, respectively.

CONCLUSIONS
Teenage girls and women who were provided contraception at no cost and educated about reversible contraception and the benefits of LARC methods had rates of pregnancy, birth, and abortion that were much lower than the national rates for sexually experienced teens. (Funded by the Susan Thompson Buffett Foundation and others.)

This study has been much-ballyhooed by contraception advocates. The study is generally well-designed, and from a cursory look, I can’t really say that the results are in any way incorrect. However, I can point out a few items that put the study results into a larger research context.

First is the somewhat obvious point that comparing the pregnancy, birth, and abortion rates for the study participants with those of “national rates among sexually experienced teens” is in some sense stacking the deck. Teens included in the CHOICE study were provided with free contraception, information about the reduced risk of pregnancy associated with IUDs, injections, etc. The average sexually active 14-19 year old nationwide is, as a result, much less likely to be actually using one of these methods of birth control. In particular, IUDs generally run about $1000, so this contraceptive method is especially unlikely to be used by the average sexually active teenage girl nationally (something contraception advocates want to change – and one of the major drivers behind the HHS mandate). As a result, the comparison of CHOICE study participants with nationwide averages for sexually active teens is really a biased comparison.

Second, the study appears to have no real control group. Yet an opportunity for such a control group was obvious. In another publication describes the counselling component of the study:

We developed a standardized contraceptive counseling script which was presented to the participant at her enrollment appointment at our university site, regardless of her baseline contraceptive knowledge or her interest in specific contraceptive methods. Participants underwent contraceptive counseling at the beginning of the enrollment process prior to completing informed consent or the baseline questionnaire. The script concisely describes the effectiveness, advantages, and disadvantages of each reversible method in order of effectiveness, including the levonorgestrel intrauterine system (LNG-IUS), the copper IUD, the subdermal implant, depot medroxyprogesterone acetate (DMPA), oral contraceptive pills (OCPs), the transdermal patch, the contraceptive vaginal ring, and condoms. Other methods such as diaphragm, contraceptive sponge, and natural family planning were discussed at the woman’s request. We provided participants with physical models of the methods during counseling and descriptions of the insertion procedures for the LARC methods. Counselors encouraged participants to ask questions and actively worked to increase accurate knowledge and dispel any incorrect information or myths.

Using this (biased) counselling, a control group could have been randomly assigned to receive free contraceptives or not. The control group could have received counseling on its own, allowing the intervention to evaluate the specific effect of the provision of free contraceptives. However, the study didn’t do that.

Aside: the NFP discussions among most academics claims a failure rate of 24%, based on lumping all methods together. I’m guessing this study did the same. The best long-term clinical studies of the sympothermal method assign it a failure rate comparable to the IUD, statistically speaking.

 

[fnr]

There’s another interesting feature of the study. While pregnancy abortion rates were significantly lower than nationally sexually-active teens, the likelihood that a pregnancy ended in abortion was higher than the national average. Nationally, the “abortion ratio” (abortions per 1000 live births) among 15-19 year olds was 332 in the year 2011, according to CDC’s latest report. In the CHOICE study, the abortion ratios for 15-19 year olds was 500 to 502. That’s quite interesting. It seems that the reduced abortion rates derive primarily from reduced pregnancy rates among study participants, but the reduced effect was offset by a greater likelihood of those who did get pregnant to get an abortion.

Also, using the data in their Figure 1, the pregnancy rates for US sexually-experienced teens and CHOICE broke out like this, with the associated %reduction as follows:
Whites: US rate 37.9, CHOICE rate 26.9 → 28.9% reduction in pregnancy rates.
Blacks: US rate 99.5, CHOICE rate 31.8 → 68.0% reduction in pregnancy rates.

These reductions are hard to evaluate, because it’s hard to know the typical contraceptive use pattern of sexually-active teens – though studies do exist.

The effectiveness of the methods in CHOICE are much, much lower than the rates published by the Guttmacher Institute on its web site (and CDC and others).

Here’s list of failure rates (in pregnancies per 100 woman-years for sexually active women) for the different types of contraceptives used in the CHOICE study as reported by Guttmacher:

Method / Perfect / Typical / Cohort percent
Hormonal IUD / 0.2 / 0.2 / 31.70%
Nonhormonal IUD / 0.6 / 0.8 / 5.30%
Implant / 0.05 / 0.05 / 34.50%
DMPA Injection / 0.2 / 6 / 9%
Oral pill / 0.3 / 9 / 12.50%
Ring / 0.3 / 9 / 4.90%
Patch / 0.3 / 9 / 2.00%

Using the composition of contraceptives in the CHOICE cohort (Table 1 of the article) to come up with an average failure rate predicted with Guttmacher rates, if the methods were being perfectly used, pregnancy rate should be 0.2 per hundred / 2 per thousand. With typical failure rates, the pregancy rate in CHOICE should have been 2.4 per hundred / 24 per thousand. Yet consistently, the pregnancy rates in CHOICE were higher than either of these estimates.