This essay is a commentary on a Research Note by James Trussell, Barbara Vaughn and Joseph Stanford entitled "Are All Contraceptive Failures Unintended Pregnancies? Evidence from the 1995 National Survey of Family Growth."

The findings of Trussell, Vaughan and Stanford raise serious questions for public health researchers who are seeking information about factors that affect maternal and child health outcomes. In trying to understand how women's pregnancy attitudes affect their well-being and that of their children, we have relied on measures of intention status as a proxy for feelings about pregnancy. Studies have linked unintended pregnancy to behaviors such as inadequate use of prenatal care or smoking and drinking during pregnancy and to negative birth outcomes such as low birthweight.1 Yet if one-quarter of women with contraceptive failures classified as unintended pregnancies are happy or very happy to be pregnant, we may need to consider what survey questions on pregnancy intention are actually measuring, and whether the responses provide a useful proxy for pregnancy attitudes.

Several issues undermine the validity of intendedness as it is currently measured. First is the recall bias implicit in retrospective questions about intention. Once a woman has a baby, she may be more likely to say that the pregnancy occurred at the right time, regardless of how she felt when she became pregnant. And perhaps, in retrospect, it did occur at the right time, even though the conception may not have been intended. Many couples with contraceptive failures can probably relate to thinking, "What a good time to have a baby."

Second, because pregnancy intention involves human emotional and psychological factors, it is an extremely complex concept. Contraceptive failure is due in large part to failure to use a method consistently or effectively, and contraceptive risk-taking may reflect ambivalence about becoming pregnant. For example, a woman may unconsciously wish to become pregnant to validate her sexuality or to secure a relationship commitment from her partner.2 Women's ambivalence about pregnancy and the unconscious wish to become pregnant—even if one doesn't really want a baby—are powerful forces that complicate the whole issue of intendedness.

The fact that two people are involved adds to this complexity. The reasons couples practice contraception when it is a good time to have a baby are too complicated to examine here, but the partners may have different views on timing. Perhaps contraceptive use reflects the man's desire to prevent pregnancy, while the woman surveyed feels the pregnancy occurred "at about the right time." Further research is warranted on couple concordance in contraceptive knowledge, attitudes and practice, including how the relative power dynamics of the relationship affect couples' contraceptive practice and pregnancy decision-making.

Discrepancies between women's stated and actual contraceptive practice also complicate the issue of intention status. As a counselor in an abortion clinic in the 1970s, I noted that most patients stated that they had been practicing contraception when they became pregnant. However, upon questioning, they usually revealed that they had not always been effective contraceptive users. Ineffective use ranged from "forgetting" to put in a diaphragm, missing one or more pills or not taking pills at the same time each day to throwing out the pills because a woman was mad at her boyfriend or because they had broken up. Because these women were terminating their pregnancies, one could be fairly confident that the pregnancies were unwanted—but were they unintended?

The issue of mistimed versus unwanted pregnancy, as measured in the National Survey of Family Growth (NSFG) and other national surveys, presents yet another problem. Unintended pregnancies are those classified as mistimed or unwanted. A pregnancy is categorized as mistimed if the woman said that she did not want a pregnancy at that time but wanted a baby at some time in the future. Pregnancies are considered unwanted if the woman said she did not want to become pregnant at that time or at any time in the future. So, the pregnancy of a young, unmarried woman having an abortion would be classified as being mistimed (but wanted) if she said she hoped to marry and have children someday. On the other hand, an accidental pregnancy to an older, married women who considered her family complete would be classified as unwanted. In fact, the opposite may be true: The young woman's "mistimed" pregnancy may truly be unwanted, while the older woman's "unwanted" pregnancy may be welcomed and carried to term.

In research, we attempt to simplify complex issues in order to quantify them. However, intendedness may be too complex to quantify and thus measure accurately. Perhaps we should drop this variable from public health research altogether, and focus instead on factors that more accurately define the issues of concern—unhappiness, ambivalence and pregnancy denial.

In 1995, the NSFG for the first time included a 10-point scale measuring the degree of happiness that a woman felt about being pregnant, as well as questions designed to measure pregnancy ambivalence. Piccinino and Peterson found consistency between the attitudinal scales and the traditional measure.3 In my own work, however, I have found that factors such as unhappiness about being pregnant and pregnancy denial were associated with use of prenatal care and low birth weight, while the traditional measure was not.4 Pregnancy denial, which has been the most robust predictor of inadequate prenatal care and low birthweight in my research, is also included in the pregnancy ambivalence measure in the latest NSFG.

Trussell, Vaughan and Stanford's findings add to the growing body of literature that challenges the usefulness of pregnancy intendedness as a measure for understanding public health outcomes. The National Center for Health Statistics has taken an important step by adding questions designed to measure pregnancy attitudes and ambivalence to the 1995 NSFG. In seeking to identify the conditions that lead to outcomes such as abortion, lack of prenatal care, substance abuse, low birth weight, lack of infant attachment, low educational attainment, poverty and violence, it makes sense to use those questions to focus specifically on women with negative attitudes toward their pregnancy, rather than on those with a pregnancy that is simply unintended.