Ninth Annual Maternal and Child Health Epidemiology Conference / December 10-12, 2003
JOHN SANTELLI: Okay, thank you. I’d like to first acknowledge my co-authors on this, Brian Morrow and Marion Carter. Adolescent pregnancy is frequently unintended. Adolescent’s use of contraception is a critical factor influencing risk of a pregnancy and a non-intended pregnancy. We have national goals for increasing use of condoms, use of contraception, including use of (inaudible) protection. Well there’ve been enormous changes in adolescent contraceptive use over the last 30 years. In the ‘70’s we saw declining use of the pill in the ‘70’s and the ‘80’s. It probably peaked actually in the ‘60’s. We don’t have good data from that period. We saw very little change in condom use in the ‘70’s, but then this dramatic increase in the ‘80’s, probably as a consequence of HIV epidemic. We had two questions we wanted to look at. What were the changes in contraceptive use among high school students between ’91 and 2001, and how did the changes in contraceptive use vary by grade, gender, and race ethnicity. We used data from the Youth Risk Behavior Survey on contraceptive use patterns among sexually active teens, those that have been sexually active in the last three months, and we created a summary measure of contraceptive effectiveness which is measured as contraceptive failure rates. We used data from the National Survey of Family Growth on contraceptive failure rates to calculate that.
Does everybody know the Youth Risk Behavior Survey? Most of you do. It’s a National Probability Survey of high school students. It’s been conducted every other year since ’91. It’s probably the best series at this point to look at what’s going on in the ‘90’s because the other measure of the National Survey of Family Growth was last conducted in ’95, so their data’s getting relatively old. The behavioral data we looked at was essentially contraceptive use at last intercourse and condom use at last intercourse, and then you can then use those two to calculate dual use. Okay. Well what’s a contraceptive failure rate? Mathematically it’s the number of pregnancies occurring within one year among 100 women using a contraceptive method, using a specific method. Today, I’m going to be talking about what’s called typical use failure rates, sort of, in the hands of the average woman, what’s the failure rate. There’s also something called theoretical effectiveness, which is in the best hands show how effective the method is.
We use published contraceptive failure rates from a study that AGI did and it’s based on the 1988 and ’95 NSFG. Well, failure rates vary pretty enormously, this gives you some sense of failure rates, so the failure rate at the top for no method is actually--the estimate is 85 women out of 100 are going to get pregnant within a year. Contrast that with--I just saw a pointer here, let me see if I can do that--this 0.54 use of injection and condom. But you can see, again, different methods vary pretty enormously. Well this is how we--we calculated something called the weighted average contraceptive failure rate. Again, it’s just a way of summarizing all these different methods and their effectiveness, and basically, you add up the method, specific contraceptive failure rate, multiplied by the percent of young people that are using that method. And then we did regression analysis to estimate the slope of change over time, and we did a *Taylor series expansion to look at annual rates of change. Let me give you some results.
These are estimated pregnancy rates among sexually active high school males and females. This weighted average contraceptive failure rate for females in 2000 was 24 pregnancies out of every 100 sexually active women. Now, we also calculated something called the weighted average contraceptive failure rate for men. What’s the problem with that? It’s real obvious. What? Men don’t get pregnant. So there’s no substantive meaning to a weighted contraceptive failure rate for men because it’s never been applied to men or calculated for men, should I say. But it actually is useful; again, still a useful summary index for what’s going on with men, and you can calculate the same changes over time. So, let me show you the slopes. What we found--and these are for males or females--you can see the weighted contraceptive failure rate, and I’ll get the numbers here--for women decline at about 14% over 10 years, or an annual rate of about 1.5%.
You can also see in this slide that the slope of the line for men was not significant. Okay. This breaks it out by race ethnicity and the only group that--you can see that the slopes are tending down but the only group for which there was significant improvement--again, the downward slope was improvement--was for black not Hispanic. And this is broken out now by grade and what you’ll see is the only group that’s really significantly showing improvement are the ninth graders, the youngest kids. And then this slide is to give you some sense of the methods that kids are using and what’s going into this index, if you will. You can see on the top line that pill use declined through ’97, may have turned around but we don’t--that’s not a statistically significant increase. You can see the condom use increase particularly in the early ‘90’s and then it seemed to have plateaued since then. Dual use on the next line--there’s two pieces to dual use, pill and condom, and pill injection on the seventh line. You can see that both of those have tended to go up over time. Importantly, there’s been a big decline in use of withdrawal, and withdrawal if you’ll remember back to the early slide has a fairly high failure rate, and decline in the use of that is important. There’s been new use in ’99. The (inaudible) began measuring use of Depo-Provera injection, so you can see new use there. And what else. And the use of no method, importantly, has decline as well from 17.6 to 14, so I think those are the major trends, if you will. We found similar changes for men, although they weren’t--as I said before--they weren’t statistically significant.
So let me summarize, pregnancy among sexually active teens is pretty common. The improvement we see in the ‘90’s was related to increased condom use, hormonal use, and in fact, if you add up pill and injection both being hormonal, we find that pretty much hormonal use in ’91 and 2001 was about the same adding those two together. There were important decreases in use of withdrawal, some small increases in dual use, but again, the levels are pretty low, and again, there’s important for unintended pregnancies there tends to be this persistent group of non-users. The limitations of the study--(inaudible) is an excellent time series and it fits the period of decline, for instance, in pregnancy rates, but it has pretty limited data (inaudible), and so we have two questions, for instance, on contraception. It doesn’t provide any information on contraceptive use for dropouts and obviously its self reported data has those limitations.
Implications, I think the improvements that we’re seeing, particularly for young women, is related to improved motivation to avoid pregnancy and STD’s and fits in with other sorts of data that we’re seeing some objective signs of that with declining pregnancy rates in the ‘90’s. It suggest very little evidence of increased access to hormonal contraception. Overwhelming the methods that kids are using are the condom and we probably do more with the clinic system to improve access for teens. There’s a persistent group of non-users, again I haven’t figured out the exact contribution of the non-users to the failure rate but it’s going to be fairly considerable. And the last thing I leave you with is stay tuned, the 2003 (inaudible) data will be out in the spring or the summer and so will the 2002 National Survey of Family Growth, which will give another look at this data, will be out next summer. Thank you.