An Exploration of the Dimensions of Pregnancy Intentions Among Women Choosing to Terminate Pregnancy or to Initiate Prenatal Care in New Orleans, Louisiana

Abstract

Objectives. We examined pregnancy decisionmaking among women seeking abortion or prenatal care.

Methods. Conventional measures of pregnancy intentions were compared with newer measures in 1017 women seeking abortion. A reduced sample of abortion patients (142 African American women from New Orleans) was compared with 464 similar women entering prenatal care.

Results. Virtually all abortion patients reported the pregnancy as unintended; two thirds of prenatal patients reported the pregnancy as unintended. Reasons for seeking abortion related to life circumstances, including cost, readiness, not wanting any more children, marital status, relationship stability, and being too young. Abortion patients were more likely to report trying hard to avoid a pregnancy and not being in a relationship. They were less likely to report that their partner wanted a baby (odds ratio=0.10) or that they wanted a baby with their partner (odds ratio=0.13) than prenatal patients.

Conclusions. Traditional measures of pregnancy intentions did not readily predict a woman’s choice to continue or abort the pregnancy. Relationship with male partners, desire for a baby with the partner, and life circumstances were critical dimensions in pregnancy decisionmaking.

Unintended pregnancies include births that are unwanted and mistimed and pregnancies that end in abortion.1^,2 Almost half (49%) of pregnancies in the United States in 1994 were unintended and almost half of unintended pregnancies end in abortion.1 Longitudinal European studies of women who had their abortion requests denied have found severe negative effects on the children’s long-term psychosocial development including effects on schooling, social adjustment, alcohol and drug abuse, criminal activity, and employment.3^,4 Where abortion is illegal and unsafe, unintended pregnancy is a major contributor to maternal morbidity and mortality.5^,6

Although the unfavorable consequences of unintended pregnancy are well delineated, unintended pregnancy itself is less well defined. Previous research suggested that pregnancy intentions are multidimensional, ambivalence about pregnancy is common, and the decision to carry a pregnancy to term or to abort is affected by life circumstances and social influences.7^–12 The complicated circumstances surrounding pregnancy decisionmaking have led to efforts to better measure pregnancy intentions.13^–17

A woman’s feelings about a specific pregnancy and her decision about abortion may be shaped by her relationship with her partner, medical conditions, social pressure from family members, physical abuse, emotional reactions, and prenatal diagnostic procedures.18^–24 Decisions about abortion are also driven by whether a woman accepts or rejects abortion as a solution for unwanted pregnancy.18^,25 Pregnancy intentions are seldom measured for women seeking abortion. A number of studies have directly compared abortion and prenatal patients on demographic and contraceptive factors;24^–26 however, these studies did not directly examine differences in pregnancy intentions between women who carry to term and women who abort.

This study used data from the Determinants of Unintended Pregnancy Risk in New Orleans Study to answer 2 questions: What are the important cognitive, affective, and contextual dimensions of pregnancy intentions among women seeking abortion? How do abortion and prenatal patients differ in terms of these cognitive, affective, and contextual dimensions of pregnancy intentions? In this study we explored traditional demographic measures of pregnancy intentions as well as new measures that were developed in a previous qualitative research phase that involved African American women from New Orleans.

METHODS

Study Design

The Determinants of Unintended Pregnancy Risk in New Orleans Study collected data from women in several clinical settings. Quantitative data were collected from women in a suburban New Orleans abortion clinic that also served inner-city New Orleans residents. This sample provided a rich perspective on abortion clients in the New Orleans metropolitan area. All women completed a 45-question self-administered survey on the day of the procedure. The surveys were placed on clipboards and given to the women to complete in the waiting area. Women read an informed consent statement about the study on the first page of the survey. Signature on the form was waived to enhance the anonymity of data collection. All completed and blank surveys were placed in a locked box in the room where the women filled out the survey. A total of 1017 completed surveys (and 6 blank surveys) were collected in 5 months (June through October 2002). The clinic averaged about 40 abortions per week during the study period, indicating that most women completed a questionnaire.

Quantitative data were also collected from currently pregnant women visiting an inner-city, free-care prenatal screening clinic. All women with a confirmed positive pregnancy test who wanted to receive free prenatal care were required to visit this screening clinic before beginning regular prenatal care. In this clinic, we used an interviewer-administered computer-assisted survey instrument. A total of 671 women completed the survey between March 2002 and February 2003. Demographic information on the number of women who refused to participate was not collected. To assess representation, we used the clinic log-book data to compare the age of women interviewed with the age of all women who visited the clinic during the same period. No significant age differences were found between interviewed women and the pool of eligible women.17

All consent procedures and survey tools were approved by the institutional review boards at Tulane University Health Sciences Center, the Centers for Disease Control and Prevention, and the oversight boards for the participating clinics.

Analysis

To answer the first research question, we developed descriptive analyses based on data from all women in the abortion clinic sample. To answer the second research question, we limited the abortion and prenatal clinic samples. Because most women in the prenatal clinic were African American (89%) and from a New Orleans zip code (74% from zip codes 70111–70196), we selected African American women with a New Orleans zip code for the comparison analysis. The reduced abortion clinic sample included 142 women. The reduced prenatal clinic sample included 464 women.

Chi-square and t tests of significance were performed across these 2 reduced samples. We also performed Fisher exact tests of differences between the 2 groups; these exact tests confirmed all significant χ² differences and are not presented. Multivariate logistic regression analyses were performed to examine which dimensions were significantly associated with choosing to abort or intending to carry to term, controlling for the sociodemographic factors. The dependent variable was coded 1 if the woman was getting an abortion and 0 if she was attending prenatal care.

To examine the cognitive, affective, and contextual dimensions of pregnancy intentions, the surveys from both clinics included questions on attitudes toward the current pregnancy. The cognitive measures included whether the pregnancy was intended, mistimed, or unwanted; how hard the woman was trying to avoid the pregnancy (on a scale of 1=not trying to avoid to 5=really trying to avoid); and whether the woman planned to get pregnant (yes/no). The affective questions were 1-to-5 scale measures that determined how happy, surprised, confused, and scared the woman was when she discovered that she was pregnant. Each of these scales was coded 1 for the low response (e.g., not happy) to 5 for the high response (e.g., very happy).

The contextual measures included whether the woman thought the pregnancy would improve her relationship with her partner (1 = not at all to 5 = a tremendous amount), whether she wanted a baby with her partner at the time (yes/no), and whether her partner wanted a baby with her (yes/no). For all of the 1-to-5 scales, the extreme values were defined and the middle categories (2, 3, and 4) were left undefined. Details of the source of the questions and question wording are presented elsewhere.17 In the abortion clinic, questions on the reasons for the abortion and the emotions of the partner on learning of the pregnancy were also asked. The partner emotion questions were asked as yes/no questions for each of the emotions included.

RESULTS

Characteristics of Women Getting an Abortion

Table 1 ▶ provides the demographic characteristics of the full abortion clinic sample, reduced abortion clinic sample, and reduced prenatal clinic sample. Among all abortion patients, 47% were White, 42% were African American, and 12% were Hispanic or another race/ethnicity. Twenty-three percent of patients lived in the city of New Orleans and 51% lived near or within 60 miles of the clinic, but not in New Orleans. Women in the reduced abortion clinic sample (all from New Orleans and African American) were similar to the larger, nonincluded sample (n = 875) on most demographic and intention dimensions examined, but were less likely to be married or engaged, more likely to be in a long-term relationship, and more likely to be very religious.

TABLE 1—

Demographic Characteristics of Women in Analysis Samples, 2002

	Full Abortion Sample Populations,a % (n = 1017)	Reduced Abortion Sample Populations,b % (n = 142)	Reduced Prenatal Clinic Sample Populations,c % (n = 464)
Race/ethnicity
African American	41.9	100.0	100.0
White	46.6	0.0	0.0
Hispanic/other	11.5	0.0	0.0
Residential location (based on reported zip code)
New Orleans	23.4	100.0	100.0
≤ 60 miles outside New Orleans	50.7	0.0	0.0
> 60 miles outside New Orleans	25.9	0.0	0.0
Age, years
< 20	17.4	19.0	27.4
20–24	36.8	41.6	40.3
25–29	22.8	23.2	20.5
≥ 30	23.0	16.2	11.9
Relationship status
Married/engaged	21.1	12.9	18.7
Relationship for ≥ 2 years	23.8	35.9	44.4
Relationship for < 2 years	28.4	21.4	25.9
No relationship	26.8	29.8	11.0**
Religiosity
Not religious at all	15.1	13.1	13.8
Somewhat religious	69.4	62.8	50.2
Very religious	14.5	23.4	28.2
Extremely religious	1.0	0.7	7.8*
Education level
< 12th grade	15.3	9.2	36.0
High school/GED completed	26.5	29.6	26.5
> High school	58.2	61.3	37.5**
Employment: Currently employedd	58.6	61.7	45.0**
Number of pregnancies (includes current)
1	19.9	11.3	29.3
2–3	40.2	43.7	47.6
≥ 4	24.1	26.8	23.1
Missing parity data	15.8	18.3	0.0**

Note: GED = general equivalency diploma.

^a Women in the greater New Orleans metropolitan area seeking an abortion.

^bAfrican American women from New Orleans seeking an abortion.

^cAfrican American women from New Orleans in prenatal care.

^dEmployment question was asked differently on the 2 surveys.

*P < .01; **P < .001 (χ² and t test significance testing compared reduced abortion sample to prenatal sample).

When we compared the reduced abortion and prenatal clinic samples, we found important differences in many sociodemographic characteristics (Table 1 ▶). Women obtaining an abortion were less likely to be currently in a relationship and to be religious and more likely to report higher educational attainment and employment than were women attending the prenatal clinic. Women in the prenatal clinic were more likely to be reporting about a first pregnancy than were women in the abortion clinic, although abortion patients seemed to have more difficulty answering the question without the assistance of an interviewer.

Multiple dimensions of pregnancy intentions, including cognitive, affective, and contextual factors, are presented in Table 2 ▶. Only 3% of the full abortion sample reported that their pregnancy was intended. Most of the abortion patients reported that the pregnancy was either mistimed or unwanted. Three percent of abortion patients reported that they planned to get pregnant at the time of pregnancy, 38% reported that they were trying hard to avoid pregnancy (code 5), and two fifths reported using contraception at the time of conception—half of this contraception was condom use (data not shown in table).

TABLE 2—

Cognitive, Affective, and Contextual Dimensions of Pregnancy Decisionmaking By Women in the Analysis Samples, 2002

	Full Abortion Sample Populationsa (n = 1017)	Reduced Abortion Sample Populationsb (n = 142)	Reduced Prenatal Clinic Sample Populationsc (n = 464)
Cognitive Dimensions
Intendedness of current pregnancy, %
Right time/later (intended)	2.7	0.7	35.2
Mistimed	51.0	56.5	29.9
Unwanted	46.3	42.8	35.0***
How hard trying to avoid pregnancy, mean, 1–5 scale	3.72	3.79	2.71***
Planned to get pregnant, %	3.2	0.7	21.7***
Affective Dimensions
Happy, mean, 1–5 scale	1.92	1.99	3.18***
Surprised, mean, 1–5 scale	3.89	3.64	3.98*
Confused, mean, 1–5 scale	3.64	3.36	2.65***
Scared, mean, 1–5 scale	3.93	3.60	2.71***
Contextual Dimensions
Would improve relationship, mean, 1–5 scale	1.56	1.67	2.08**
Wanted baby with partner, %	6.6	3.7	50.9***
Partner wanted baby with you, %	17.6	12.9	68.7***

Note: Percentages may not sum to 100% because of rounding.

^a Women in the greater New Orleans metropolitan area seeking an abortion.

^bAfrican American women from New Orleans seeking an abortion.

^cAfrican American women from New Orleans in prenatal care.

*P ≤ .05; **P ≤ .01; ***P ≤ .001 (χ² and t test significance testing compared reduced abortion sample to prenatal sample).

Women in the study were asked to describe their initial emotional reaction to the pregnancy on 4 affective dimensions: happy, surprised, confused, and scared. Among women in the full abortion clinic sample, more than half reported that they were very unhappy about the pregnancy. Only 5% reported being very happy about the pregnancy (code 5). The women in the abortion clinic sample also reported being very surprised, very confused, and very scared about the pregnancy (mean values higher than 3.5).

The women were also asked whether their partners were surprised, confused, shocked, scared, happy, upset, or angry when they learned about the current pregnancy (with yes/no questions). The women who were getting an abortion reported that their partners experienced similar emotional reactions, including being surprised (37%), confused (35%), and shocked (33%). Notably, 20% of the women reported that their partners were happy about the pregnancy compared with only 9% of the women who reported a 4 or 5 on the happiness scale. Fourteen percent of the women reported that the partners were not informed about the pregnancy and the abortion decision; half of these women were not currently in a relationship.

Contextual dimensions are also presented in Table 2 ▶. Only 7% of the full abortion sample reported that they wanted a baby with their partner at the time. A larger percentage of women reported that their partners wanted a baby with them at the time (18%). The mean value on the expectation that the pregnancy would improve the relationship with their partner was low; few women expected the pregnancy to improve their relationship.

Figure 1 ▶ illustrates the reasons that women reported they were getting an abortion (multiple responses were possible). The primary reason given by almost half of the women was that they could not afford a child. Many responses were related to pregnancy intendedness (either mistimed, unwanted, or wanted now); 40% reported not being ready for a child (mistimed pregnancy) and 36% reported not wanting any more children (unwanted pregnancy). Additionally, the response “too young” probably reflects mistimed pregnancies and the response “too old” probably reflects unwanted pregnancies. Four of the reasons reflect relationship issues: not married, relationship was unstable, terminated the relationship, and marital or legal problems. Health reasons for getting an abortion included problems with the woman’s own health and concerns about the health of the baby.

FIGURE 1—

Most common reasons for abortion, full abortion clinic sample (N = 1017), 2002.

Contextual reasons for getting an abortion varied by age. Among women aged younger than 20 years, the most common reasons for obtaining an abortion were too young (73%), not ready for a child (64%), can’t afford a child (56%), not married (45%), and relationship unstable (20%). Among women aged 30 years or older, the most common reasons were: don’t want anymore children (57%), can’t afford a child (33%), too old (26%), not ready for a child (22%), and relationship unstable (20%).

Table 2 ▶ compares the reduced abortion sample and the reduced prenatal sample. In the reduced abortion sample, only 1 woman reported that her pregnancy was intended; a third of women in the prenatal sample reported that their pregnancy was intended. Of pregnancies that were unintended, a greater proportion in the abortion clinic was mistimed whereas in the prenatal clinic, a greater proportion was unwanted. Abortion patients had a significantly greater mean value on how hard they were trying to avoid pregnancy. Only 1 abortion patient planned her pregnancy compared with 22% of the prenatal patients.

The 2 clinic populations also showed significant differences in the expected directions in 3 of 4 initial emotional reactions to the pregnancy (Table 2 ▶). Abortion patients reported being less happy (lower mean value), more confused (higher mean value), and more scared about the pregnancy than were prenatal patients. Additionally, the mean value on how surprised the women were at the time of pregnancy detection is significantly higher for prenatal patients than for abortion patients. This finding is counterintuitive given that abortion patients were significantly more likely than prenatal patients to report using contraception at the time of pregnancy. Our previous work with this population has indicated that there are misperceptions about the risk of pregnancy that also might explain this result.11^,27

The mean value on the perception that the pregnancy would improve the relationship was significantly higher in the prenatal clinic than in the abortion clinic. Abortion patients were substantially less likely to report that they wanted a baby with their partner and that their partner wanted a baby with them compared with prenatal patients.

Multivariate Analyses

Multivariate analyses were conducted to compare abortion patients (coded 1) with prenatal patients (coded 0). Model 1 in Table 3 ▶ includes only the sociodemographic factors, Model 2 includes only the dimensions of pregnancy intentions, and Model 3 includes all variables. In Models 2 and 3, the conventional intendedness measure (intended vs mistimed and unwanted) and the planning status variable were not included because, in the abortion clinic sample, only 1 woman intended and 1 woman planned her pregnancy, making it difficult to model these variables.

TABLE 3—

Odds Ratios (95% Confidence Intervals) of Demographic and Psychosocial Factors Associated With Having an Abortion Compared With Carrying to Term

	Model 1: Demographic	Model 2: Intentions	Model 3: Demographic +Intentions
Demographic factors
Age, years
< 20 (reference)	1.00		1.00
20–24	0.92 (0.48, 1.78)		0.99 (0.42, 2.33)
25–29	1.05 (0.49, 2.24)		0.98 (0.35, 2.70)
≥30	1.34 (0.57, 3.13)		1.53 (0.49, 4.78)
Relationship status
Married/engaged (reference)	1.00		1.00
Relationship for ≥ 2 years	1.70 (0.81, 3.56)		1.09 (0.43, 2.76)
Relationship for < 2 years	2.68* (1.19, 6.01)		1.60 (0.57, 4.49)
No relationship	7.38*** (3.25, 16.79)		3.19* (1.13, 8.98)
Religiosity
Not religious at all	2.75** (1.27, 5.97)		1.69 (0.64, 4.50)
Somewhat religious	2.06* (1.18, 3.58)		1.64 (0.82, 3.28)
Very to extremely religious (reference)	1.00		1.00
Education level
< 12th grade (reference)	1.00		1.00
High school/GED completed	4.85*** (2.08, 11.29)		7.77*** (2.79, 21.67)
> High school	8.49*** (3.81, 18.96)		9.81*** (3.76, 25.56)
Employed	1.74* (1.07, 2.83)		2.11* (1.14, 3.90)
Cognitive dimension
How hard trying to avoid pregnancy (1–5)		1.29* (1.06, 1.57)	1.32* (1.05, 1.65)
Affective dimensions
How happy about pregnancy (1–5)		0.87 (0.70, 1.10)	0.91 (0.70, 1.17)
How surprised about pregnancy (1–5)		0.73** (0.60, 0.89)	0.66*** (0.52, 0.84)
How confused about pregnancy (1–5)		1.07 (0.87, 1.31)	1.14 (0.90, 1.44)
How scared about pregnancy (1–5)		1.17 (0.97, 1.42)	1.16 (0.93, 1.44)
Contextual dimensions
How much pregnancy would improve relationship (1–5)		0.97 (0.78, 1.21)	1.01 (0.79, 1.29)
Wanted baby with partner
No (reference)		1.00	1.00
Yes		0.12*** (0.03, 0.41)	0.13** (0.03, 0.47)
Partner wanted baby
No (reference)		1.00	1.00
Yes		0.11*** (0.05, 0.22)	0.10*** (0.05, 0.22)
Statistical Testing
Model: N	543	543	543
Likelihood ratio χ2 (df)	86.01 (11)	193.28 (8)	246.53 (19)
Pseudo R2	0.1580	0.3549	0.4527

Notes: GED = general equivalency diploma. Could not estimate model with conventional intendedness (intended, mistimed, unwanted) and planning status in the model because there was little variability in these measures in the abortion clinic (1 intended pregnancy and 1 planned pregnancy). All 1–5 measures are entered as continuous variables in the models presented.

*P ≤ .05; **P ≤ .01; ***P ≤ .001; based on multivariate logistic modeling.

In Model 1, the important sociodemographic differences between the 2 groups were relationship status, religiosity, education, and employment. Women who were not currently in a relationship were much more likely to be getting an abortion than women who were married or engaged (odds ratio [OR] = 7.4). Women who were in a relationship that had lasted less than 2 years (most were ≤ 1 year) were also more likely to get an abortion than women who were married or engaged (OR = 2.7). No differences between groups were found between women in long-term (≥2 years) relationships and women who were married or engaged. The remaining demographic variables were in the expected directions.

In Model 2, women who wanted a baby with their partner were less likely to be seeking an abortion (OR = 0.12) than women who did not want a baby with their partner. Likewise, women who reported that their partner wanted a baby with them were less likely to be getting an abortion (OR = 0.11). Women who were more surprised about the pregnancy were also less likely to be getting an abortion. Finally, women who were trying harder to avoid a pregnancy were more likely to be getting an abortion. In Model 2, none of the other intention dimensions differentiated between abortion and prenatal patients.

In Model 3, the intention measures had the same pattern of effects, such that the partner variables “trying to avoid” and “surprised” remained significantly associated with the decision to abort versus to carry to term. The odds ratios for these factors were not diminished by adjustment with the sociodemographic variables. In the full model the relationship status and religiosity effects were attenuated somewhat whereas education and employment effects were not.

DISCUSSION

Virtually all abortion patients in our sample from New Orleans reported that the pregnancy was unintended; however, two thirds of prenatal patients also reported that the pregnancy was unintended. The women reported various reasons for seeking an abortion and most related to life circumstances, including being unable to afford a child, not ready for a child, not wanting any more children, not married, too young, and having an unstable relationship. Abortion patients were less likely than prenatal patients to report being surprised about the pregnancy. The relationship with male partners was an important influence in shaping women’s decisions about abortion; women reported being unmarried or relationship instability as reasons for obtaining an abortion. Perceived partner intentions and desire for a baby with the partner were strong predictors of the decision to choose abortion over childbearing; in fact, these 2 questions best discriminated between abortion and prenatal patients.

Traditional measures of pregnancy intentions did not readily predict between these patient groups in the bivariate analyses. As such, most women reporting a pregnancy as intended on the traditional measure will continue the pregnancy, whereas women reporting the pregnancy as not intended may end up choosing to continue or terminate the pregnancy, presumably on the basis of other considerations.

Our findings parallel previous research on the importance of relationship status and male partner influence on a woman’s decision to seek abortion.18^,19^,26^,28^–33 Previous research demonstrated that abortion patients were more likely to be unmarried.31 We also found that specific questions about male partners and their intentions attenuated the influence of relationship status as defined by marriage and length of the relationship. This statistical effect should be expected because these aspects of relationship status are correlated. Similar to previous work, our study also showed that education and employment remained strong influences on the likelihood of seeking abortion even after we controlled for pregnancy intentions.31 This should not be surprising because having a child can significantly affect educational plans and employment opportunities.

We found that women reported a variety of emotional reactions when discovering that they were pregnant. Previous research with the National Survey of Family Growth suggested that emotional reactions, particularly happiness about the pregnancy, are important in understanding a woman’s attitude toward her pregnancy.14 In our multivariate modeling, women who were more surprised about the pregnancy were more likely to be carrying to term. This may reflect misperceptions of the pregnancy risk among these women.

The contextual circumstances that women in our study reported for seeking abortion were also similar to those found previously.24^,28^,32^,34 Bankole and colleagues28 summarized studies from 27 diverse countries, including the United States, and found the most common reasons for having an abortion were (in order) the desire to postpone or stop child-bearing, employment and educational concerns, the lack of support from partners, and the need to care for existing children. Sihvo and colleagues34 found that reasons for having an abortion varied considerably by the age of the woman. All these findings reinforce the importance of life circumstances of women in shaping the abortion decision and understanding pregnancy intentions.

Limitations

Our sample, although at high risk for unintended pregnancy, represents a primarily inner-city population from a single city. These results cannot be directly generalized to other populations.

In New Orleans, it was not easy to find an abortion clinic that was similar in terms of sociodemographics to the prenatal clinic. As such, we had to create a reduced abortion sample that was more comparable to our prenatal sample. In 1996, Louisiana only had 15 abortion providers and 92% of women in the state were living in localities without an abortion provider.35 Therefore, it is not surprising that the abortion clinic covered a wide distribution of women. Thus, our study had limited power when comparing the reduced abortion sample with African American prenatal patients from New Orleans. Moreover, a key expected difference—income—was not measured in this study so we were unable to directly control for this difference and instead used proxies such as education and employment status.

We did not assess partner intentions directly; our assessments of these intentions were based on the women’s understanding of men’s feelings and preferences. In reporting partners’ reactions, women may project their own feelings.

Finally, the surveys in the 2 clinics were administered differently. Overall, there were more missing data in the abortion clinic sample, particularly for the question on parity.

Implications for Public Health Research and Practice

When assessing and counseling women about family planning or pregnancy decision-making, reproductive health practitioners should explore the broad range of contextual, affective, and cognitive dimensions that influence the women’s choices. Exploring partner dimensions may also be useful to clinicians counseling women at the time of pregnancy testing. The strength of women’s motivations as reflected in questions about how hard a woman is trying to avoid pregnancy, her expected emotional reaction if she were to become pregnant, and the perceived support of her partner may also be useful in helping women to improve contraceptive adherence.

Given multiple, often conflicted, desires and social influences, most women choosing abortion in this study seemed to be making rational and responsible decisions. The women in this study reported a variety of reasons for choosing abortion, particularly social contextual factors such as not being able to afford a child, being too young, not being ready to be a mother, and not being in a stable relationship. These reasons suggest that the women choosing abortion valued childbearing but also understood its important implications for themselves and potential offspring. The women in this analysis did not seem to be making the decision about abortion lightly or impulsively. As such, clinicians should be respectful and supportive of women’s choices.

Many pregnancies ending in abortion resulted from relationships that were reported by the women to be weak. Encouraging couples within new relationships to delay sexual intercourse and to be effective contraceptive users may help prevent unintended pregnancy and thereby reduce recourse to abortion. Family planning and primary care programs also need to reach out to young men in promoting reproductive health—for themselves and their partners. Providing a financially supportive social context to strengthen relationships between young men and women, such as employment opportunities and reasonably priced housing for young families, may also strengthen the ability of couples to keep and raise children.

Our study also has implications for measuring and interpreting pregnancy intentions. We found that the conventional measure of whether a pregnancy is mistimed or unwanted did not discriminate well between women who decided to abort and women who decided to carry to term. Other research has suggested that the conventional demographic questions on pregnancy intentions work well at a population level in assessing the prevalence of pregnancy intentions and in estimating unmet need for contraception but that they work less well in assessing the feelings and needs of individual women.7^,12 A recent paper from our group suggested that pregnancy desirability, which combines questions on a woman’s feelings, plans, and relationship, may better assess pregnancy intentions in demographic surveys.17

These analyses suggest the need for a better understanding of women’s pregnancy intentions and the multiple influences on women’s decisions to continue or abort a pregnancy. Such an understanding may ultimately contribute to improved prevention of unintended pregnancy and, thus, reduce recourse to abortion.