Advancing Sexual and Reproductive Health and Rights
Family Planning Perspectives
Volume 28, Number 6, November/December 1996

The Family Planning Attitudes and Experiences of Low-Income Women

By Jacqueline Darroch Forrest and Jennifer J. Frost

A 1995 telephone survey of 1,852 low-income women aged 18-34 who were sexually active and at risk of unintended pregnancy found that 83% were currently practicing contraception. They were more likely to do so if they held positive attitudes toward contraceptive use, if they talked frequently about intimate matters with their partners and girlfriends and if they were very satisfied with the services they received at their last gynecologic visit. Seventy percent of current users said they were very satisfied with their method. Women whose last visit was to a clinic, who were very satisfied with the care they received and who used the pill or a long-acting method were more likely than others to report being very satisfied with their contraceptive. Women very satisfied with their gynecologic care were more likely to use oral contraceptives and to take them consistently, but were less likely to report that their partner used condoms or, if they did, used them consistently. Most women had made a medical visit for gynecologic or contraceptive care in the past year (86%), and 80% were very satisfied overall with their care at their last visit. Women were more likely to be very satisfied if the staff was courteous, helpful and respectful and made an effort to find out their needs, if their clinician's gender matched their own preference and if the facility was clean and services were organized.

(Family Planning Perspectives, 28:246-255 & 277, 1996)

The majority of American women experience an unintended pregnancy at least once in their lives,1 and more than half of all pregnancies are unplanned. Low-income and minority women have greater difficulties than other women in avoiding unplanned pregnancy: Seventy-four percent of pregnancies to women with a family income less than 150% of the federal poverty level are unplanned, compared with 52% of those among higher income women. Likewise, 79% of pregnancies among black women are unintended, compared with 63% of those among Hispanic women and 54% of those among white and other non-Hispanic women.2 The greater prevalence of unplanned pregnancy among low-income women and among black or Hispanic women reflects their lower levels of contraceptive use3 and higher likelihood of contraceptive failure.4 However, it is not clear whether racial and ethnic differences simply represent the greater socioeconomic disadvantage of black and Hispanic women or if they result from other characteristics of or barriers faced by these women.

In this article, we seek to learn directly from low-income women about their perceptions of and experiences with contraceptive methods and contraceptive service providers. To help identify characteristics that might be risk factors for not using a method, for being dissatisfied with a method or for using one inconsistently, we explore the effects of women's personal characteristics, their relationships, their childbearing experience and plans, their motivation to avoid unplanned pregnancy and their attitudes about contraception. We also examine women's experiences with and evaluation of contraceptive service providers. Because women of various racial and ethnic backgrounds have different levels of unplanned pregnancy, we give special attention throughout this article to identifying similarities and differences between white, black and Hispanic women.


A nationally representative sample of 1,852 low-income women was interviewed by telephone between January and April 1995.* Respondents were eligible for the study if their family income was below 200% of the federal poverty level, if they were white, black or Hispanic women aged 18-34, and if they were at risk of an unintended pregnancy. We oversampled blacks and Hispanics, yielding a total of 1,852 respondents—454 white, 451 black and 947 Hispanic women.§ Trained interviewers conducted the 25-minute interviews in Spanish or English; most interviewers were matched according to the respondent's race or ethnicity. Women were told that the interview was part of a national study about the health care needs of women—one that would be used to plan better health care services. They were also informed that some personal or sensitive questions regarding reproductive health and contraceptive use would be asked during the interview.

To obtain the sample of low-income women in a cost-efficient manner, we limited telephone calls to the 40% of exchanges (the area code and first three digits of phone numbers) in which at least 25% of all households had an income below $15,000. These exchanges covered 33% of all U.S. households.

To increase the likelihood of contacting black and Hispanic respondents, the selected exchanges were divided into three independent strata: exchanges in which at least 25% of the households were Hispanic, exchanges in which at least 25% of the households were black and exchanges that fit neither of the first two categories. A fourth stratum was formed from the numbers in the latter exchange that were found on a list of Spanish-surname telephone numbers. Random digit samples of telephone numbers were generated from each of the first three strata and a random sample was pulled from the list of Hispanic-surname numbers. Survey participants were screened and recruited as they were identified in any of the sampling strata, regardless of their race or ethnicity, until the study sample was roughly 50% Hispanic, 25% black and 25% white women.

Overall, 91,939 telephone numbers were dialed as part of the screening process.** A total of 2,054 women in the screened households were eligible for the study. Of these, 90% agreed to participate and completed the questionnaire—94% of eligible women in Hispanic-surname households, 92% of women in the higher-density black areas, 89% of those in the higher-density Hispanic areas and 86% of women in the remaining stratum.

In the analyses reported here, race and ethnicity reflect the woman's self-report rather than the type of sample stratum in which she lived. Hispanic respondents who said they predominantly used Spanish in their daily life were classified as Spanish-speaking; all others were classified as English-speaking. This classification was used as an indicator of the extent to which these respondents were native to or acculturated to the predominantly English-speaking U.S. culture, or were closely tied to the U.S. Hispanic community or to their native country.

Roughly equal proportions of the English-speaking Hispanic women reported that their family had come from either the United States or Mexico (38% and 36%, respectively); 17% had come from Puerto Rico and 3% were from the Dominican Republic or Cuba. Seventy-two percent of English-speaking Hispanic women were born in the United States; 17% had come to the United States more than 15 years before, and 11% had come more recently. Almost all Spanish-speaking Hispanic women (97%) had come from another country; 35% had come to the United States between 1990 and 1995 and 52% had come between 1980 and 1989. Most of these women were originally from Mexico (74%), with 6% from the Dominican Republic, 5% from El Salvador, 4% from Puerto Rico and the remaining 11% from other countries.

The data presented in this article have been weighted to take into account the differing probabilities for each stratum that the telephone number of an eligible women would actually be dialed, differences in response rates across strata, and estimated differences within strata of response rates, according to race and ethnicity. The distributions therefore represent national estimates of the characteristics of low-income women who reside in relatively low-income neighborhoods and who are currently at risk of an unintended pregnancy.

In this article, we first explore bivariate differences by racial and ethnic subgroup in a number of relevant variables, such as personal characteristics, women's relationships with their partners, their childbearing history, their experiences with different types of service providers and their attitudes about and experiences with contraceptive use. (All bivariate racial and ethnic differences discussed here are significant at p<.05.*) We then performed logistic regression analyses to control for the effects of these variables. For these regressions, we adjusted the weights so that the total number of cases in the analysis equaled the unweighted total number of respondents, by dividing the weight originally assigned to each respondent by the ratio of the total number of weighted cases to the total unweighted number.*


Personal Variables

Sixty-three percent of the respondents were white, 20% were black and 17% were Hispanic—7% English-speaking and 10% Spanish-speaking (Table 1, page 248). More than half of respondents were younger than 25; 21% did not have a high school diploma, while only 41% had attended college. Forty-one percent were currently working, and another 28% were in school. Respondents were fairly evenly spread across four categories of religious affiliation. Forty percent reported being covered by private insurance over the past year, 30% were covered by Medicaid or another form of public insurance and 30% had no coverage.

Table 1. Percentage distribution of low-income women aged 18-34 and at risk of unintended pregnancy, by selected characteristics, according to race and ethnicity, 1995, AGI Survey of Low-Income Women
Characteristic Total White Black Hispanic
(N=1,852) (N=454) (N=451) Total English-speaking Spanish-speaking
(N=947) (N=364) (N=583)
% distribution†100.062.820.117.06.910.1
18-1915.113.624.210.117.4 5.2
20-2438.741.633.1 34.443.927.9
25-2926.126.722.5 27.821.432.3
<high school20.814.217.848.625.564.5
High school diploma‡38.639.842.229.942.621.2
Some college/college grad40.646.0 39.921.531.914.3
School/employment status
In school§27.826.839.018.332.78.4
At home31.528.723.551.133.263.5
Roman Catholic25.115.911.275.473.476.8
Fundamentalist Protestant26.426.739.99.48.510.0
Other Protestant23.126.828.
Other/none/don't know25.330.620.411.912.611.4
Medical coverage
Private insurance40.147.836.716.124.210.5
No coverage29.529.415.146.632.656.3
Marital status
Never married33.528.260.721.134.811.7
Length of time with partner
No steady partner14.813.021.013.919.610.0
<1 year14.915.115.513.717.611.0
1-2 years24.426.025.217.318.416.6
>=3 years45.945.938.355.144.362.4
Frequency of sex††
No answer2.
0 33.839.428.419.531.211.4
Future childbearing
Want more59.464.
Want no more32.327.346.134.330.137.2
Don't know8.38.55.810.58.911.7
Feelings if pregnant now
Very unhappy20.818.633.614.318.511.2
A little unhappy22.222.026.617.823.114.4
Don't care16.417.916.011.410.811.8
A little glad20.921.915.223.821.525.7
Very glad19.719.68.732.726.236.9
†Weighted. ‡Includes respondents who received the General Educational Development (GED) high school equivalency certificate. §Includes respondents who were both in school and working. ††Frequent is two times per week or more; moderate is once per week, on average, or a few times per month; and infrequent is once per month or less. Note: Percentages may not sum to 100.0% because of rounding.

Respondents tended to differ by race and ethnicity on many of these characteristics. Blacks were younger and Spanish-speaking Hispanics were older than other women. Levels of completed education were lowest among Hispanics, especially among Spanish-speaking women. Black women were most likely to be currently in school, while Spanish-speaking women were most likely to be at home (i.e., neither in school nor working).

Hispanics were predominantly Catholic, while blacks were most likely to say they were fundamentalist Protestants and whites were most likely to be of some other religion or to report no religious affiliation. White women were most likely to be covered by private insurance, while Medicaid was the most common type of health coverage among black women and English-speaking Hispanics; Spanish-speaking women were most likely to have had no medical coverage in the past year.


More than half of the respondents (57%) were currently married or living with an unmarried partner. Another 9% were separated, divorced or widowed, and 34% had never been married. While 15% had no steady partner, 46% had been with their current partner for at least three years.

Black women were most likely to be unmarried, and Spanish-speaking Hispanic women were most likely to be currently married. Black women were also more likely than whites and Spanish-speaking women to have no steady partner, and Spanish-speaking women were more likely than others to have been with their partner for three years or more.

Almost half of respondents (48%) had intercourse frequently (two or more times per week) during the preceding three months, 29% had sex moderately often (once per week or a few times per month) and 20% had sex infrequently (once per month or less).*§ Frequency of intercourse was highest among cohabiting or married women, with 68% and 58%, respectively, reporting that they had sex at least twice a week (not shown). Half (51%) of those who were divorced, separated or widowed had sex once a month or less, as did one-third of never-married women.

Childbearing Variables

Seventy-two percent of respondents had been pregnant at least once, ranging from 67% of whites to 90% of Spanish-speaking women* (not shown). While 66% had had at least one child, most had had only one (31%) or two (22%).

Fifty-nine percent wanted to have a child at some future time, even though none were currently trying to become pregnant. Fifty percent of those who wanted to have a child in the future had no children, 33% had one child, 12% had two children and 5% had three or more (not shown).

Overall, 32% of women said they did not want to have children in the future. Almost two-thirds of these women already had two (34%) or more (28%) children, while 28% wanted to stop childbearing after one child and 11% planned to remain childless (not shown). Whites were most likely to have had no children and least likely to have had three or more. White women and English-speaking Hispanics were more likely than were other women to want additional children.

Almost all respondents (99%) gave at least one response to an open-ended question asking why they did not want to become pregnant or to have a child at that time, and 30% gave two or more. Twenty-nine percent said they could not afford to have a child, and 19% said they already had as many children as or more than they wanted. Other reasons given by women for not wanting to become pregnant at that time were that they had a baby or a young child (16%), that they were in school (14%), that they were not married (12%), that they were too young (6%), that they were not ready to have a child at that time (6%), that they could not have a child because of work or other goals (5%) or that they had health-related reasons (3%).

However, in response to the next survey question, 20% said they would feel very glad if they found out they were pregnant, 21% said they would feel a little glad and similar proportions said they would feel a little unhappy or very unhappy. A smaller proportion did not care either way or did not know how they would feel. Black women were most likely to say they would be very unhappy and least likely to be very glad. Spanish-speaking women were most likely to say they would be very glad.

In a logistic regression that controlled for personal and relationship characteristics and for childbearing experiences and plans, the difference between blacks and whites was not statistically significant. But whether they spoke English or Spanish, Hispanic women were significantly more likely than white women to say they would feel very glad if they found out they were pregnant.

The logistic regression also indicated that women who were 20 or older, who were not working or going to school, who were fundamentalist Protestant, who had been with their partner for less than a year or who wanted children in the future were more likely than others to say they would feel very glad if they found out that they were pregnant. Subgroups less likely to say they would be very glad included women who had attended college, unmarried women, those who had sex moderately often, women with children and those whose reason for not wanting a child at the time was economic or because of having a baby or young child at home.


Bivariate Relationships

Nearly all of the women said they had made a gynecologic or contraceptive visit at some time in the past five years, and most had done so in the past year (Table 2). Seventy-six percent of all women had a Pap test in the past year, 61% had a breast exam, 57% obtained a contraceptive method, 31% had a pregnancy test, 13% made a visit related to a problem that they were having with their method and 6% obtained diagnosis or treatment for a sexually transmitted disease (not shown). Black women were the most likely to have made a visit in the last year and Hispanic women were least likely to have done so.

Table 2. Percentage of low-income women aged 18-34 and at risk of unintended pregnancy, by provider and contraceptive experiences, according to race and ethnicity
Experiences Total White Black Hispanic
(N=1,852) (N=454) (N=451) Total English-speaking Spanish-speaking
(N=947) (N=364) (N=583)
% who made gynecologic visit
In past 5 years94.995.396.791.293.189.8
In past year85.585.791.677.680.775.5
Type of provider at last visit†
Private MD/HMO49.254.845.432.243.224.2
Hospital clinic11.18.716.313.916.112.3
Health department clinic15.512.820.819.113.922.8
Planned Parenthood/family planning clinic17.215.615.026.421.929.7
Other clinic7.
Amount paid for last visit†
Clinician's sex†
Client's preferred sex35.
Not client's preferred sex9.67.58.618.820.018.0
Does not matter55.255.562.245.348.343.1
Experiences at last visit†,‡
Waiting rooms are not crowded39.144.234.924.634.017.8
One feels comfortable there66.372.658.551.556.947.6
Services are organized70.678.454.660.665.557.1
Facility is clean93.094.588.093.594.293.0
Staff finds out needs69.075.964.547.455.441.6
Staff is courteous80.582.078.876.976.777.0
Staff is respectful87.589.
One receives quality care84.987.480.980.279.780.7
Summary rating of experience†
Very positive§46.456.633.621.331.513.6
Satisfaction with services†
Very satisfied80.482.978.473.475.172.2
Somewhat satisfied14.512.717.518.017.218.5
Neutral/not satisfied5.
Talked more than once a month about contraception
With partner44.341.946.150.644.754.7
With girlfriends50.650.357.743.248.939.2
With mother18.016.722.817.221.014.5
Contraceptive attitudes‡
Contraception matters in pregnancy timing54.860.053.137.547.430.7
Don't need partner approval of contraception69.
Contraceptive use doesn't spoil sex72.077.872.450.660.244.0
Friends think contraception is important70.170.269.470.869.371.8
Most effective method currently used
Any method83.485.584.474.674.474.7
Long-acting methods12.510.016.017.713.020.9
Oral contraceptives39.443.835.028.325.530.3
No method16.614.515.625.425.625.3
Had sex recently12.310.511.919.117.120.4
Did not have sex recently4.
Multiple methods33.032.646.419.024.215.1
Condoms plus other27.726.143.315.020.111.5
Satisfaction with contraceptive††
% very satisfied70.171.870.862.365.360.0
†Among those making a visit in past five years. ‡The proportion given Is the percentage that completely agreed with the statement. §The percentage with a very positive rating represents the proportion having a total score of 36 or more out of a possible 40 on the eight experience variables. ††Among current users, for the most effective method used.

Roughly half of the women who had made a contraceptive or gynecologic visit in the last five years had most recently visited a clinic, and half had gone to a private doctor or health maintenance organization (HMO).†† Women who visited clinics most often went to health department clinics (16%) or Planned Parenthood and similar family planning clinics (17%). Smaller proportions had visited a hospital clinic or some other type of facility. White women were the most likely to have visited a private doctor or HMO, while Spanish-speaking women were most likely to have gone to a clinic for their last visit.

Overall, 43% of respondents who had made a gynecologic or contraceptive visit in the past year had received services for free, either because the costs were totally covered by insurance or Medicaid or because the women went to a provider that offered free care to low-income people. This represented 38% of those who had last been to a private physician or HMO and 48% of women who had been to a clinic.

Another 25% of women—33% of those who had gone to a private provider and 17% of those who went to a clinic—paid for some of the cost of their last visit. These included 18% whose visit was partially paid for by private insurance and 7% who had Medicaid coverage. Women whose visit costs were partially covered paid a median of $25 per visit, typically $20 at a private provider and $30 at a clinic.

Thirty-two percent of women received no assistance in paying for their visit. Some 29% of women who went to a private provider and 35% of women using a clinic paid a median of $48—$60 at private providers and $40 at clinics. (It is not known whether the amount women paid represented the full charges or the amount they were able to pay, nor whether the amount they were charged was a reduced fee, as is typical for low-income women in family planning clinics.) Black women and English-speaking Hispanic women were most likely to have paid nothing at their last visit, while whites were most likely to have paid $20 or more.

Although 55% of women said that the sex of their clinician did not matter, 37% preferred to have their gynecologic exam performed by a woman, while 8% preferred a male practitioner. Among the 45% with a preference, 35% had a clinician of the desired gender at their last visit and 10% did not. Ninety percent of women said they had no preference regarding the clinician's race or ethnicity (not shown); 7% were served by a clinician of their preferred race or ethnicity and 3% were not. Eighty-two percent of white women were served by a white clinician, but 68-82% of other women were served by someone who was not of their race or ethnicity.

Almost all (96%) of the Spanish-speaking women and 13% of English-speaking Hispanics preferred to be served by Spanish-speaking providers. Only 69% of those who preferred Spanish-speaking service providers reported that a doctor (41%) or a nurse (27%) spoke Spanish with them at their last visit. Another 12% reported that a receptionist spoke Spanish, but 19% said no one spoke Spanish at their last visit.

Women were asked whether they completely agreed, somewhat agreed, were neutral, somewhat disagreed or completely disagreed with each of eight statements describing the staff and facility at their last gynecologic visit.†‡ While only 39% reported that the waiting rooms were not too crowded, most respondents gave otherwise positive assessments: Two-thirds or more said that they felt comfortable waiting with the other people using that provider, that services were organized and that the staff made an effort to find out their needs. More than 80% judged the staff to be courteous and helpful, said they were treated with respect and felt they received quality care, and nearly all perceived the rooms and equipment to be clean.

Forty-six percent of respondents rated their experience very highly, with a summary score of 36 or more.†§ White women were most likely to rate the staff and facility very highly (57%) and Spanish-speaking women were the least likely to do so (14%).

The very low proportion of Spanish-speaking women who assigned a high rating to their last visit reflected, in part, a language barrier. Among all Hispanic women who preferred an English-speaking provider, 33% gave a summary rating of 36 or more for their last visit; in comparison, among those who preferred to receive care in Spanish and had someone (a doctor, nurse or receptionist) speak Spanish, 15% rated their visit highly, while among those who preferred Spanish but encountered no one who spoke Spanish, only 7% did so.

In response to a direct question, 80% of women said that overall they were very satisfied with the care they received at their last visit; whites were more likely to have been very satisfied than were black or Hispanic women.


Any gynecologic visit. After using logistic regression to control for the effects of personal characteristics, we found that black women remained significantly more likely than white women to have made a visit for gynecologic or contraceptive care in the past year, while Hispanics were significantly less likely to have done so (Table 3). Women covered by Medicaid or other public insurance were much more likely than those with private insurance to have made a visit, but there was no significant difference between women with private insurance and those with no coverage. Other groups of women more likely to have made a visit in the past year included those in college and those who would be a little unhappy if they became pregnant. Older women, women who were neither Catholic nor fundamentalist Protestant and those who had sex infrequently were significantly less likely to have made a recent gynecologic visit.

Table 3. Odds ratios from logistic regressions showing effects of selected variables on use of gynecologic services and satisfaction with them
Variables Made gyn. visit in past year Last visit was to a clinic† Positive rating of last visit† Very satisfied with care†
(N=1,813) (N=1,702) (N=1,611) (N=1,623)
Hispanic, English-speaking0.50*1.78*0.51*1.09
Hispanic, Spanish-speaking0.43**5.34***0.22***0.93
<high school1.
High school grad/GED0.911.101.140.66
Some college/college grad1.93**0.71*2.06***1.15
Roman Catholic1.
Fundamentalist Protestant1.071.221.401.77*
Other Protestant0.53**1.70**1.77**1.63
Other/none/don't know0.58*1.92***0.941.03
Medical coverage
Private insurance1.
No coverage0.892.13***0.74*0.83
Marital status
Never married0.760.860.61**1.15
Frequency of sex
No answer0.861.261.191.32
Future childbearing
Want no more1.
Don't know1.301.460.57*1.59
Want more1.260.800.751.19
Feelings if pregnant now
Very unhappy1.
A little unhappy1.91**1.011.80**0.87
Don't care0.721.56*1.55*1.45
A little glad1.
Very glad1.101.301.79**1.31
Gynecologic provider
Private MD/HMOnana1.001.00
Hospital clinicnana0.66*0.59*
Health department clinicnana0.45***0.42***
Planned Parenthood/family planning clinicnana1.071.45
Other clinicnana0.40***0.95
Amount paid at last visit
No visit/missing datanana0.740.30**
Clinician's sex
Does not matternana1.001.00
Client's preferred sexnana0.961.49*
Not client's preferred sexnana0.33***0.45***
Experiences at last visit
Waiting rooms were not crowdednanana1.32
One feels comfortable therenanana0.97
Services are organizednanana2.48***
Facility is cleannanana1.98*
Staff finds out needsnanana1.71**
Staff is courteousnanana3.10***
Staff is respectfulnanana5.61***
*In this and later tables, association is statistically significant at p<.05. **Association is statistically significant at p<.01. ***Association is statistically significant at p<.001. †Among women who made a gynecologic visit in the the past five years. Notes: Other variables included in the logistic regressions but not significant were school/employment status, length of time with partner and clinician's race/ethnicity. In this and later tables, Ns vary because of nonresponse on different variables. na=not applicable.

Type of gynecologic provider. With other factors controlled, the odds of having gone to a clinic for gynecologic care were significantly higher for both black women and Hispanic women, particularly for Spanish-speaking women, than they were for white women. Those with Medicaid coverage or no medical coverage were also significantly more likely to have been to a clinic, as were women who were neither Catholic nor fundamentalist Protestant, those who had sex infrequently and those who said that they would not mind becoming pregnant. Women aged 25-34, those who had attended college and those with 1-2 children were less likely than were comparison groups of women to have used a clinic at their last visit.

Rating of visit. We found white women to be more likely than the others to give their experience at their last gynecologic visit a very positive rating, using the eight statements that described the staff and facility. Spanish-speaking women were least likely to do so.

Women who had been to college, nonfundamentalist Protestants and women who had sex infrequently were also more likely to have rated the staff and facility very highly, as were women who would not be very unhappy if they became pregnant. Women whose last visit was to a Planned Parenthood or family planning clinic were as likely as those who went to a private physician to give it a high rating, but women who went to a hospital, a health department or another type of clinic were significantly less likely to do so, as were women with no health care coverage, never-married women, women with 1-2 children and those unsure if they wanted children in the future. Women who preferred a clinician of a particular sex but were served by someone of the opposite sex were less likely to give their last visit a very positive rating.

Satisfaction with care. While there were few differences by women's background characteristics in whether women were very satisfied overall with the services they received, some provider-related characteristics were significant predictors of overall satisfaction with services. Women reporting that the staff was courteous and helpful, that the staff made an effort to find out their needs and were respectful, and that the facility was clean and services organized were significantly more likely than those who did not to report being very satisfied. Women who paid $20 or more for their visit were less likely than those who received free care or whose care was paid for entirely by insurance or Medicaid to be satisfied. Those served by a clinician of the preferred sex were significantly more likely than those with no preference to be very satisfied, while those served by a clinician who was not of their preferred sex were less likely.

Independent of these factors, women who went to health department or hospital clinics were significantly less likely to be very satisfied with their care than were women who went to private physicians. When provider-related variables were controlled, cohabiting women were less likely than married women to be very satisfied, and fundamentalist Protestants and those who had infrequent intercourse were more likely to be very satisfied.


Bivariate Relationships

Forty-four percent of respondents discussed contraception with their partner at least once a month, and 51% said they talked once a month or more with at least one close girlfriend about such personal matters as sexual relationships and contraceptive use (Table 2). Only 18% of respondents talked at least once a month with their mother about sex or contraception. (In part, this is because 19% reported that their mother was dead or lived too far away to communicate with regularly. This proportion reached 39% among Spanish-speaking women.)

Most respondents had a favorable attitude toward contraceptive use, but a substantial minority did not. Only 55% of respondents completely disagreed with the statement, "It doesn't matter whether I use birth control or not; when it is my time to get pregnant, it will happen." Sixty-nine percent completely disagreed with the statement, "I would not use birth control if my partner disapproved;" 72% completely disagreed that "Planning ahead about birth control can spoil the fun of sex;" and 70% completely agreed that their "... friends believe it is important to use birth control when you have sex."

If the above replies are considered the most conducive to contraceptive use, then only 26% of women completely identified with all four. This proportion ranged from 29-30% of white and black women to 21% of English-speaking Hispanics and 6% of Spanish-speaking women (not shown).

Eighty-three percent of respondents said they were currently using a contraceptive method; most (74%) of those not using a method had had sex in the last three months. White women and black women were more likely than Hispanic women to be practicing contraception.

When method use was classified according to the most effective method women reported,‡* 39% of all respondents were currently using the pill and 13% were using long-acting reversible methods such as injectables (9%), the implant (2%) or the IUD (1%). Twenty-four percent used male condoms, and 7% relied on other barrier methods or on periodic abstinence and withdrawal.

White contraceptive users appeared more likely than others to be using the pill, while Spanish-speaking contraceptive users were most likely to be relying on long-acting methods. Black women were most likely to use condoms, although the difference between them and English-speaking Hispanics was not statistically significant.

Overall, 33% of women reported using more than one contraceptive method, and 28% said they used condoms along with another method, with black women the most likely to report double method use (Table 2). Thirty-two percent of women using long-acting methods and 38% of pill users also used condoms, often for the purpose of preventing infection (not shown). Almost one-third (32%) of all women who reported current use of male condoms said they were using the method primarily to prevent pregnancy, while 6% did so primarily to prevent infection; the majority (63%) reported having both goals in mind. Twenty-nine percent of married condom users said that preventing infection was one of their reasons for using condoms, compared with 69% of cohabiting, 85% of formerly married and 95% of never-married women.

Women using contraceptives reported overwhelmingly that they were satisfied with their current method—70% very satisfied and 19% somewhat satisfied. White and black contraceptive users were more likely to report being very satisfied with their method than were members of either Hispanic subgroup.

Many respondents reported they did not use their method consistently (not shown). Only 44% of pill users said that in the past three months they had never missed a day of pill use; 44% had forgotten to take their pills for 1-2 days and 12% for three or more days in a month. Of those relying on condoms, 38% said that their partner used them at every act of intercourse in the last three months, 41% said condoms were used most or about half of the time and 20% said they were used occasionally or not at all. White pill users were more likely to report never having missed a pill (48%) than were blacks (33%) or English-speaking Hispanics (36%).


Contraceptive use. In logistic regressions that excluded nonusers who had not had sex in the past three months (Table 4), only English-speaking Hispanic women were significantly less likely than whites to be using a contraceptive method. Those who had been to college, cohabiting and never-married women, women in their relationship for a year or more, those with 1-2 children and women who wanted children in the future had an elevated likelihood of using contraceptives. The odds of contraceptive use among women who would be very glad if they became pregnant were about half those of women who would be very unhappy if they became pregnant.

Table 4. Odds ratios from logistic regressions showing effect of selected variables on method use
Variable Any use† Pill‡ Long-acting‡ Condoms‡,§
White 1.00 1.00 1.00 1.00
Black 0.74 0.67* 1.32 1.87***
Hispanic, English-speaking 0.51* 0.49** 1.89 1.75*
Hispanic, Spanish-speaking 0.70 0.80 4.76*** 0.94
18-19 1.00 1.00 1.00 1.00
20-24 0.86 1.82** 0.51** 0.87
25-29 0.94 1.70* 0.25*** 1.00
30-34 0.65 1.61 0.26*** 1.42
<high school grad. 1.00 1.00 1.00 1.00
High school grad/GED 1.42 0.77 1.46 1.41
Some college/college grad 3.42*** 0.72 2.01** 1.01
In school 1.00 1.00 1.00 1.00
Working 0.84 0.87 2.88*** 0.79
At home 1.01 0.51*** 2.77*** 1.04
Roman Catholic 1.00 1.00 1.00 1.00
Fundamentalist Protestant 0.85 0.93 1.11 1.59*
Other Protestant 0.73 1.11 1.61 0.81
Other/none/don't know 0.63 0.70* 1.48 1.39
Medical coverage
Private insurance 1.00 1.00 1.00 1.00
Medicaid/public 0.73 1.02 1.55* 0.51***
None 0.68 0.82 1.00 0.87
Marital status
Married 1.00 1.00 1.00 1.00
Cohabiting 1.76* 1.09 1.78* 0.56**
Divorced/separated/widowed 1.04 1.41 1.63 0.39**
Never married 2.56** 1.12 1.18 0.55**
Length of time with partner
No steady partner 1.00 1.00 1.00 1.00
<1 year 1.71 0.88 1.60 0.51*
1-2 years 2.08* 1.29 0.83 0.45**
>=3 years 3.10** 1.48 1.30 0.36***
Frequency of sex
Frequent 1.00 1.00 1.00 1.00
Moderate 0.81 0.69** 1.04 1.68***
Infrequent 1.18 0.67* 1.62 1.59*
No answer 0.68 0.91 1.27 0.55
0 1.00 1.00 1.00 1.00
1-2 1.63* 0.63** 3.23*** 0.66*
>=3 1.22 0.40*** 2.88** 1.06
Future childbearing
Want no more 1.00 1.00 1.00 1.00
Don't know 0.87 0.61* 0.91 1.60
Want more 1.58* 1.09 0.61** 1.09
Feelings if pregnant now
Very unhappy 1.00 1.00 1.00 1.00
A little unhappy 1.16 1.21 0.99 0.81
Don't care 0.92 0.90 1.17 1.37
A little glad 1.16 1.16 1.53 0.59*
Very glad 0.48* 0.87 1.43 0.78
Gynecologic provider
Private MD/HMO 1.00 1.00 1.00 1.00
Clinic 1.38 1.47** 1.35 0.72*
No recent visit 0.70 0.23** 1.32 1.58
Satisfaction with services
Not very satisfied 1.00 1.00 1.00 1.00
Very satisfied 2.47 *** 2.04*** 1.03 0.54***
Talk >=1 times per month about contraception
With partner 1.96*** 1.04 0.51*** 1.64***
With girlfriends 1.67** 0.74* 1.07 1.34*
With mother 0.76 1.29 0.88 0.84
Contraceptive attitudes
Contraception matters in preg. timing 1.69** 0.99 1.09 1.29
Don't need partner's approval 1.59** 1.10 1.60* 0.87
Contraceptive use doesn't spoil sex 0.84 1.12 1.09 0.59***
Friends think contraception is important 1.87*** 1.24 0.90 0.89
†Regression excludes women using no method who reported no sexual intercourse during the past three months (N=1,712). ‡Regression is based on use of specified method versus use of all other methods among all current users (N=1,448). §Includes combined use of condom and less-effective methods.

Women very satisfied with the care they received were more likely to be contraceptive users than were women who were less satisfied. Women were also more likely to be using a method if they talked frequently with their partner or girlfriends about contraception, if they felt strongly that they did not need their partners' approval for contraceptive use, if they believed that contraception is important in preventing pregnancy and if they perceived that their friends see contraceptive use as important.

Pill use. Logistic regression analysis evaluating use of particular methods indicated that black women (odds ratio of 0.67) and English-speaking Hispanics (0.49) were significantly less likely to be using the pill, once differences between groups were taken into account. Pill use was more likely among women aged 20-29, those who used clinics for their gynecologic care and those very satisfied with the care they received. Pill use was less likely among women who were neither in school nor working, who did not belong to any of the major religious groups, who had sex moderately often or infrequently, who already had children, who were unsure about their future childbearing plans, who had not made a gynecologic or contraceptive visit in the last five years and who talked with a girlfriend at least once a month about personal issues.

Use of long-acting methods. Our logistic regression confirmed that Spanish-speaking contraceptive users were much more likely to be using long-acting reversible methods than were white women. Others who had an elevated likelihood of using long-acting methods included women who had been to college, those not currently in school, women covered by Medicaid, cohabiting women, those who already had children and women who felt that they did not need their partner's approval to practice contraception. Contraceptive users aged 20-34 and those who wanted to have children in the future were less likely to have selected long-acting methods, as were women who talked with their partner about personal topics like contraception at least once a month.

Condom use. When we controlled for the effects of other variables, black and English-speaking Hispanic contraceptive users were significantly more likely than white women to rely on condoms as their most effective method. Fundamentalist Protestants, those who had sex moderately often or infrequently and women who talked frequently with their partner or girlfriend about intimate subjects were also more likely to be using condoms. The likelihood of condom use was reduced among women who were covered by Medicaid, who had 1-2 children, who would feel a little glad if they became pregnant, who used clinics for their gynecologic care and who were very satisfied with the services they received.

The longer a contraceptive user had been with her partner, the lower were her odds of using condoms. Once we had controlled for whether unmarried women had a steady partner and for how long women had been with their current partner, we found that unmarried women were less likely to use condoms as their most effective method; condom use was also less likely among those who felt that planning ahead for contraception does not spoil the fun of sex.

Method satisfaction. A logistic regression analysis that examined method satisfaction showed that, even when other variables were controlled (Table 5, page 254), Spanish-speaking women remained significantly less likely than whites to be very satisfied with their contraceptive.

Table 5. Odds ratios from logistic regressions showing effect of selected variables on satisfaction with and consistency of contraceptive use
Variables Very satisfied with method†
Never missed a day using pill‡
Used condom every time§
Hispanic, English-speaking0.811.021.28
Hispanic, Spanish-speaking0.46**2.59*0.78
<high school grad1.001.001.00
High school grad/GED0.961.121.53
Some college/college grad1.001.88*1.20
Roman Catholic1.001.001.00
Fundamentalist Protestant0.66*0.941.09
Other Protestant0.721.180.82
Other/none/don't know0.731.660.59
Marital status
Never married1.67**1.262.23*
Frequency of sex
No answer0.545.20**2.36
Future childbearing
Wants no more1.001.001.00
Don't know0.48**0.18***0.79
Wants more0.740.47**1.20
Feelings if pregnant now
Very unhappy1.001.001.00
A little unhappy1.200.57*0.86
Don't care1.503.27***0.67
A little glad1.240.820.87
Very glad1.480.990.73
Gynecologic provider
Private MD/HMO1.001.001.00
No recent visit2.23*0.481.37
Satisfaction with services
Not very satisfied1.001.001.00
Very satisfied1.79***1.99*0.42***
Talk >=1 times per month about contraception
With partner0.801.210.57*
With girlfriends1.160.851.53*
With mother0.71*0.57*2.60***
Contraceptive attitudes
Contraception matters in pregnancy timing1.040.700.63*
Don't need partner's approval of contraception0.69*0.921.16
Contraceptive use doesn't spoil sex2.02***4.07***1.06
Friends think contraception is important1.171.491.47
Contraceptive method used
Condoms alone1.18na3.39***
Oral contraceptives3.07*** na0.95
Long-acting methods2.52***na0.82
Reason for using condoms
Pregnancy preventionnana1.00
Disease preventionnana3.48*
†Includes all women currently using any method. ‡Includes all women currently using oral contraceptives. §Includes all women currently using condoms. Note: Other variables included in the logistic regressions but not significant were school/employment status, medical coverage and length of time with partner.

Women were also less likely to be very satisfied with their method if they were fundamentalist Protestant, if they were not sure whether they wanted a child in the future, if they talked frequently with their mother about sexual matters and if they felt they did not need their partner's approval for contraceptive use. Those who were more likely to feel very satisfied with their method included those aged 25-29, never-married women, those who had been to a clinic at their last visit or who had made no recent visit, and those who were very satisfied with the care they received, as well as women who felt that planning ahead for contraception does not spoil the fun of sex and women using the pill or long-acting methods.

Consistency of pill use. A logistic regression to determine consistency of pill use showed that black pill users remained less likely than white pill users to report having taken all of the necessary pills in the last three months, but that Spanish-speaking women were more likely than whites to report consistent use. Other groups more likely to report taking all of their pills were women who had been to college, formerly married women, those who gave no information about how often they had sex, women who would not care if they became pregnant, those who were very satisfied with care at their last visit and women who felt that planning ahead does not spoil the fun of sex.

In contrast, women aged 20-34 were less likely than those 18-19 to say they had taken all of their pills in the last three months, as were those who were unsure about future childbearing or who wanted more children, who would feel a little unhappy if they became pregnant now or who talked frequently with their mother about sex and contraception. Women who had been to a clinic for a recent visit were also less likely to report taking all of their pills.

Consistency of condom use. There were few bivariate differences by race and ethnicity in the consistency of condom use, and logistic regression showed no significant racial and ethnic differences either. Condom users aged 20 and older were significantly more likely to say they always used a condom than were younger women. Never-married condom users were more likely than currently married women to have used condoms consistently, as were those who had sex moderately often, who frequently discussed contraception with their girlfriends or mother, who used condoms as their only method and who relied on condoms solely for prevention of infection. Condom users were less likely to report consistent use if they had three or more children, relied on a clinic, were very satisfied with their most recent visit, talked with their partner frequently about contraception or thought contraceptive use was important in preventing pregnancy.


We focused our study of women's family planning-related attitudes and experiences on low-income women at risk of unintended pregnancy because, as a group, they are more likely than higher income women not to use a contraceptive and are more likely to get pregnant even when they do. Therefore, they are more likely to experience an unplanned pregnancy. We do not know, however, whether the women in our study differ from higher income women in terms of their feelings about pregnancy, attitudes about birth control and their experiences with contraceptive service providers.

The information that we obtained provides insight into the problems that some Americans have in avoiding unplanned pregnancy and raises questions for research and challenges for service delivery. Researchers are beginning to explore gradations in motivation or ambiguity among women who say they do not want to have a child at the present time. Data from our study suggest that women who have obvious reasons for not wanting a child at the time (because they are not married, have had all the children they want, have a baby or young child or cannot afford to raise a child) may be firmer in wanting to avoid childbearing.

A recent Institute of Medicine report urges the entire nation to adopt a new social norm that all pregnancies should be intended, emphasizing the importance that couples consciously plan for pregnancy.5 Women who think they would feel very glad if they became pregnant, even though they would not want to have a child at the time, are less likely to practice contraception. Providers' and women's open acknowledgment of the frequent coexistence of feelings for and against having a child could help women with mixed feelings better assess their own goals.

Most U.S. women want and have a small number of children, and they spend most of their reproductive lives trying to avoid becoming pregnant and in need of contraception.6 The high proportion of our low-income respondents who hold at least some views likely to interfere with effective contraceptive use—74%—suggests great potential for becoming pregnant unintentionally. This is a priority topic for further research, both within other groups of women and men and across countries. Indeed, ambivalence about sexuality in general and about contraception in particular has been suggested as one of the key reasons for higher levels of unintended pregnancy and abortion in the United States than exist in other industrialized Western countries.7

In our study, women who thought that contraception really matters in determining whether or not one gets pregnant and women whose friends think contraceptive use is important were significantly more likely to practice contraception, as were women who talk frequently about relationships and contraception with their girlfriends, partners and mothers. One challenge is building supportive attitudes and messages into these conversations.

Women who think that planning ahead for birth control can spoil the fun of sex are no more or less likely than others to use contraceptives, but they are less likely to be satisfied with their method and, if they use the pill, less likely to use it consistently. This finding lends support to the urgency of developing less intrusive, easier-to-use methods. It also confirms that ambivalence about sexuality can contribute to unintended pregnancy. More forthright acceptance of sexuality and help for couples o integrate their contraceptive use into their sexual relationship might therefore help reduce rates of unplanned births and abortions. While contraceptive service providers obviously have a role to play, this is a broader societal task as well. Salient public and provider messages could well have a positive impact on method use.

We sought not only to identify ways to help low-income women prevent unintended pregnancy more effectively, but also to assess barriers to success among low-income white, black and Hispanic women. Some of the racial and ethnic differences evident in simple cross-tabulations were shown in multivariate analyses to be the result of social and demographic differences between racial and ethnic groups. Those variations that remained may still reflect other unmeasured differences between the groups, such as the types of service providers available in the areas where they live. Black and Hispanic women, for example, are more likely to use clinics and less likely to report that they had a positive experience at their last visit, yet there are no significant differences in their satisfaction with their last visit. This may indicate that from experience, low-income black and Hispanic women have come to expect a lower standard of care than low-income white women. This is both an area for future research and a challenge for providers to investigate whether clients are truly pleased and comfortable with the services they receive.

Racial and ethnic differences were also seen among contraceptive users in terms of method selection, with Spanish-speaking Hispanic women more likely to use long-acting methods (primarily injectables), black and English-speaking Hispanic women more likely to use condoms and whites more likely to use the pill. Further research would help to identify the extent to which such differences reflect more positive attitudes toward specific methods in different communities, or greater suspicion of or concern about alternative methods. For instance, higher use of long-acting methods among Spanish-speakers may reflect greater experience and comfort with some of these methods in Central and South America, where many of the women lived previously. Alternatively, Spanish-speaking women may not be well-informed about other methods or their providers may lack confidence that such women can use other reversible methods effectively.

Our research clearly shows that women's experiences with contraceptive and gynecologic providers affect both whether and how well women use contraceptives. Attention to the basics of courtesy and respect and, among many Hispanic women, providing services in the client's language should be minimum criteria for service providers. In fact, the study provides strong evidence that treating clients well not only leads them to feel better about services, but actually improves their chances of controlling their fertility—the ultimate goal of clients and providers alike. Among the women surveyed, those who were very satisfied with their last contraceptive or gynecologic visit were more likely to use a contraceptive, more satisfied with the method they used, more likely to choose the pill and more apt to use it consistently.

Women who were very satisfied with services were less likely to rely on condoms and less likely to use condoms consistently. This finding may illustrate one of the costs of men's lack of contact with contraceptive counseling or education outside of school sexuality education. Providing information, education and skills-building experiences can help women influence their partners to use condoms consistently, but the inclusion of men themselves is also needed.

Improving services is likely to require more staff time and therefore greater expense, costs that need to be supported by private insurance, Medicaid and other public sources and, if possible, by the clients themselves. (More than half of these low-income women had paid something—often substantial amounts—toward their last visit.) Increasing support for services to low-income women and men will be a challenge in an era of cost-containment, especially since key areas for improvement are those that are often not reimbursed by strictly medical financing mechanisms—staffing levels and attitude, client education and counseling, and outreach through community programs to affect broader societal attitudes about pregnancy and contraception.

The information gathered from low-income women points to factors that can be changed, publicly and privately, in ways that might improve contraceptive use and increase couples' ability to have the number of children they want when they want them. Negative attitudes about contraception and about sexuality in general interfere with effective fertility control. Services provided in a manner that is not respectful and sensitive to clients' needs contribute to lower and less-effective contraceptive use.

These are not the only factors affecting method use, but they are clearly important. Furthermore, they are factors that can be altered by the actions of friends and families, providers, community organizations and the media, among others. Improvements are possible, and they can help women and their partners better achieve their own goals and can help lower our national level of unplanned pregnancy.

*Nationwide, 5% of households lack a telephone—4% of white households, 13% of black households and 12% of Hispanic-origin households. (See: U.S. Bureau of the Census, Phoneless in America, Statistical Brief 94-16, July 1994.). For all households in the exchanges sampled for this study, it was estimated that 9% lacked phones. Clearly, by using telephone interviews, we missed those women in households without phones—women who are especially likely to be poor. We assume that difficulties respondents might have had in obtaining and using contraceptives would be even greater among these women.

The cutoff of 200% was chosen because women with an income below this level are eligible for many subsidized reproductive health services.

A woman was defined as being at risk of an unintended pregnancy if she had had sex with a man in the last 12 months, if she was not currently pregnant, if she did not want to become pregnant and if neither she nor her partner was sterilized (either for contraceptive or medical reasons).

§Native American, Asian and Pacific Islander women were not included in the survey; moreover, since interviews were conducted only in English and Spanish, women who were not comfortable in either language could not be included. Hispanic respondents could be of any race. In this article, white non-Hispanic and black non-Hispanic respondents are referred to simply as white or black.

**Among the 91,939 numbers called, 31,519 were businesses or disconnected numbers. At 21,369, no one could be reached on three different attempts, because of no answer, a busy signal or an answering machine. Up to six calls were made to the remaining 38,940 numbers that could be identified as households. Respondents were fully screened for eligibility at 27,549; at 11,503, either the initial respondent or the woman in the household who potentially met inclusion criteria was unwilling, unable or unavailable to provide the necessary information.

*†The following percentage-point differences in proportions between racial and ethnic subgroups were significant at p<.05, based on one-tailed Z-tests and proportions of 50%: white vs. black (5.5); white vs. Hispanic (4.7); white vs. English-speaking Hispanic (5.8); whites vs. Spanish-speaking Hispanics (5.2); blacks vs. Hispanics (4.7); blacks vs. English-speaking Hispanics (5.8); blacks vs. Spanish-speaking Hispanics (5.2); and English-speaking Hispanics vs. Spanish-speaking Hispanics (5.5).

*‡This allows the tests of statistical significance to be based on the sampled number of respondents, but with the appropriately weighted distribution of respondents. Thus, while the tests of statistical significance in the logistic regressions are based on the unweighted number of cases, the odds ratios reflect the weighted distributions of respondents.

Included in the last category are 10% of all women who reported having had no sex in the past three months.

†*Forty-one percent of women reported having had at least one unplanned pregnancy: 49-50% of black women and English-speaking Hispanics, compared with 39% of whites and 33% of Spanish-speaking Hispanics. The number who reported that their last unplanned pregnancy ended in a birth was more than five times as high as the number who reported that it ended in abortion. National data show that roughly equal proportions of all unintended pregnancies end as births and as abortions, and that the number of unintended births among women with an income below 150% of the poverty level is only 1.2 times the number of abortions. This suggests that respondents underreported their numbers of unplanned pregnancies (see reference 2).

††Only 4.6% of all women had gone to an HMO for their most recent visit; these respondents have been combined with the women who visited a private physician.

†‡The statements read were: "The waiting rooms are often too crowded;" "I don't feel comfortable waiting with the other people who go there;" "The services there are often disorganized;" "The rooms and equipment are all clean;" "The people who work there do not make an effort to find out my needs;" "The people who work there are courteous and helpful;" "The staff treat me with respect;" and "The G-Y-N care I receive there is good quality." Coding ranged from 1 (for the most negative response) to 5 (for the most positive response).

†§Reply scores for the eight questions were summed to create an overall scale of the quality of experiences that women had in obtaining care, with values ranging from 8 (most negative) to 40 (most positive).

‡*Women reporting current use of more than one method were classified according to the most effective method that they reported, in the following order of presumed effectiveness: implant, IUD, injectable, oral contraceptive, male condom, female condom, diaphragm, spermicide, periodic abstinence and withdrawal.


1. J.D. Forrest, "Unintended Pregnancy Among American Women," Family Planning Perspectives, 19:76-77, 1987.

2. K. Kost and J.D. Forrest, "Intention Status of U. S. Births in 1988: Differences by Mothers' Socioeconomic and Demographic Characteristics," Family Planning Perspectives, 27:11-17, 1995.

3. J.D. Forrest and S. Singh, "The Sexual and Reproductive Behavior of American Women, 1982-1988," Family Planning Perspectives, 22:206-214, 1990; and W.D. Mosher and W.F. Pratt, "Contraceptive Use in the United States, 1973-88," Advance Data from the National Center for Health Statistics, No. 182, 1990.

4. E.F. Jones and J.D. Forrest, "Contraceptive Failure Rates Based on the 1988 NSFG," Family Planning Perspectives, 24:12-19, 1992.

5. S. Brown and L. Eisenberg, eds., The Best Intentions: Unintended Pregnancy and the Well-Being of Children and Families, National Academy Press, Washington, D.C., 1995.

6. J.D. Forrest, "Timing of Reproductive Life Stages," Obstetrics & Gynecology, 83:105-111, 1993.

7. E.F. Jones et al., Pregnancy, Contraception, and Family Planning Services in Industrialized Countries, Yale University Press, New Haven, Conn., 1989.

Jacqueline Darroch Forrest is senior vice president and vice president for research and Jennifer J. Frost is senior research associate, The Alan Guttmacher Institute (AGI), New York. The authors would like to thank David Michaelson and Denise Marks of Market Development, Inc., for their assistance in the fielding of this survey, as well as Theresa Camelo, Saba Zeleka, Michele Bolzan and Suzette Audam of AGI for their assistance during various phases of the project. The authors also thank the more than 20 members of the advisory committee for the study, who met several times during the course of the project and provided valuable insight and direction. The study on which this article is based was supported by the Henry J. Kaiser Family Foundation and The Robert Wood Johnson Foundation.