taking each pill within the appropriate time window, abstaining from sex or using a backup method when necessary, obtaining refills on time, stopping one cycle and starting the next at the right time, interpreting problems correctly (neither over-reacting nor under-reacting), and taking effective action to resolve problems (see Chapter 5 for more discussion of the skills required for contraceptive use). The point is that many reversible methods are difficult to use perfectly all of the time, and therefore user failure should be seen as reflecting both the skills of the user as well as the inherent complexity of many available reversible methods themselves.
Recognizing these realities, researchers studying contraceptive failure rates often distinguish between typical use and perfect use, thus providing some context for separating mechanical failure from user foible. Perfect use reflects contraceptive use that is consistently performed according to the specified instructions; all pregnancies occurring in the presence of perfect use are classified as method failures. Typical use, on the other hand, reflects a combination of actual method failure and user failure—a more real-world, everyday measure. Rates of failure are substantially higher with typical use than with perfect use.
Not surprisingly, in typical use, coitus-dependent methods are significantly less effective than coitus-independent methods, especially those that are longer-acting. For example, the first-year contraceptive failure rate for condoms and diaphragms ranges from 12 to 20 percent (Hatcher et al., 1994), and the 6-month failure rate of the female condom is estimated to be 12 percent (Trussell et al., 1994). By contrast, contraceptive implants and injections have less than a 1 percent failure rate (Ross, 1989; Hatcher et al., 1994). Overall, those reversible methods that are nearly impervious to user shortcomings are most effective in day-to-day life.
Jones and Forrest (1992) computed failure rates of several commonly used methods based on the 1982 and 1988 NSFG. These rates, shown in Table 4-4, were calculated on the basis of the first 12 months of use, and therefore may not reflect failure rates over longer periods of time. The investigators suggest that rates of contraceptive failure may have increased during the 1980s, especially for some methods. Failure rates for oral contraceptives, for example, increased from 6 to 8 percent between 1982 and 1988—a trend that is particularly worrisome given the fact that under conditions of perfect use, oral contraceptives yield only about one pregnancy in 1,000 women in the first year of use. This increase can also be seen in less reliable methods; for example, the failure rate for periodic abstinence increased from 16 to 25 percent during the 1980s.
It is important to emphasize that, given the imperfect array of contraceptives available to both men and women and the years of exposure to the risk of unintended pregnancy, some appreciable number of unintended pregnancies will inevitably occur over the life course. One simple computation suggests that this accumulates to a large risk over time. Ross (1989) offers the example of a young