Pregnancy Intention and Use of Contraception Among Hispanic Women in the United States: Data from the National Survey of Family Growth, 2006–2010

Abstract

Background

Both unintended and adolescent childbearing disproportionately impact the Hispanic population of the United States.

Methods

We used the 2006–2010 National Survey of Family Growth (NSFG) to provide the most recent, nationally representative description of pregnancy, childbearing, and contraception for Hispanic females aged 15–44. We determined baseline fertility data for self-identified Hispanic female respondents. Among those reporting a pregnancy history, we calculated the proportion of pregnancies identified as unintended and their association with sociodemographic variables. We also assessed outcomes and estimates of relative risk for unintended pregnancy. Finally, we examined contraceptive use prior to self-reported unintended pregnancies.

Results

Approximately 70% of Hispanic women reported ever being pregnant, including 18% of teenagers. Over half (51%) of those pregnancies were unintended, including 81% among teenagers. The adjusted risk of unintended pregnancy was highest in women 15 to 19 years old and those with three or more pregnancies (incidence rate ratio [IRR] 1.64, 95% confidence interval [CI]: 1.44–1.88 and IRR 1.77, 95% CI: 1.53–2.06, respectively). Half of unintended pregnancies were preceded by no contraception. The most common reason for unintended pregnancy preceded by contraception was “improper use” (45%) and among pregnancies without use, the most common response (37%) was “I did not think I could get pregnant.”

Conclusions

There is a high frequency of unintended pregnancy and lack of contraceptive use among Hispanic women. These findings highlight the need for improved reproductive education and contraceptive counseling in this population.

Introduction

According to the United States Census Bureau, 2011 was the first year in which more than half the children born in the United States were from a racial or ethnic minority.¹ Hispanics represent the largest portion of these minority births (1,049,128 births, 52%)¹ and have had the highest birth rate in the United States for over 20 years.^2,3 Between 2009 and 2050, the proportion of American children under the age of five who are Hispanic is projected to grow from 25% to 40%.^4,5

The reality remains that not all of these births are intended, and just as Hispanics have the highest birth rate, they also have the highest unintended birth rate.⁶ Data from 2006 reveal that Hispanics have more than double the unintended birth rate of non-Hispanic whites, 45 per 1000 births versus 18 per 1000.⁶ Among teenagers, a group where an estimated three out of four births are unintended,⁷ Hispanics have had the highest birth rate since 1995.^2,3,8 Hispanic women and adolescents have also been shown to be less likely to use contraception during intercourse than non-Hispanics.^9–12 Consequently, as the Hispanic population expands, this disparity in unintended and adolescent childbearing is also at risk of increasing.

Culturally tailored interventions show promise towards reducing health disparities.^13,14 There are very few, if any, reproductive health interventions developed specifically for the Hispanic population that target family planning. For example, in a large systematic review of culturally specific interventions and health disparities, no study assessed family planning interventions in Hispanics.¹³ With that in mind, we used data from the most recent, 2006–2010, National Survey of Family Growth (NSFG) to inform future initiatives for unintended pregnancy specifically developed for Hispanic women. Our study had two aims with an overarching goal of obtaining a better understanding of pregnancy intention within the Hispanic population, rather than a comparative analysis with other populations. First, we wanted to expand upon the descriptive data provided in reports from the National Survey of Family Growth and perform a detailed analysis of pregnancy intention and risk for unintended pregnancy among Hispanic American women. Second, we wanted to examine contraceptive behavior prior to unintended pregnancies in this population.

Previous studies using both primary and secondary data sources have associated various sociodemographic variables with unintended pregnancy and contraceptive use in both the general and the Hispanic population, including age, education, income, immigration status, religion, language preference and marital status.^6,7,15–20 However, to our knowledge, no other study has utilized multiple regression techniques to identify independent risk factors for unintended pregnancy in a nationally representative sample of Hispanic women. Additionally, unlike the majority of publications where rates and risk for unintended pregnancy are presented with pregnancies as the unit of analysis, we chose to assess risk where women were the unit of analysis. We chose women as the denominator since interventions target individuals, rather than pregnancies. Taken together, our findings provide detailed data regarding Hispanic women's pregnancy intention and contraceptive behavior and will benefit programs developing unintended pregnancy interventions tailored for the Hispanic population.

Materials and Methods

Data source and study design

We present data from the 2006–2010 NSFG, a survey conducted by the U.S. Centers for Disease Control and Prevention's National Center for Health Statistics (NCHS). The purpose of the NSFG is to provide nationally representative data on topics related to reproductive health. It is considered the preeminent source of data on unintended pregnancy and is used to determine unintended pregnancy targets for the Healthy People Reports.^6,21–23 The 2006–2010 survey was the seventh time it has been conducted since 1973. It consisted of in-person interviews by trained female interviewers who ask about pregnancy, sexual activity, contraception, fertility, reproductive and general health services, marital status, adoption, and attitudes about sex, pregnancy, and marriage, as well as partner information. Participants were selected from a national probability sample of men and women, aged 15–44, and the overall response rate was 77%, 80% for Hispanics.²⁴ Interviews were conducted in both English and Spanish. Hispanics, African Americans and teens were oversampled to provide better national estimates. Further details about the NSFG sampling methodology can be found in previous NCHS reports.^25,26

This was a cross sectional analysis performed in two stages and used data from two separate but matched NSFG datasets. The first dataset, the female respondent file, contains data on the 12,279 women who were interviewed. The NSFG pregnancy file contains data per pregnancy, such that each observation represents one of the 20,497 pregnancies reported by the 7538 women who reported ever being pregnant. For women who reported an unintended pregnancy within three years of the interview date, additional questions were asked regarding contraceptive use prior to conception.

Inclusion criteria

We included all women in the NSFG who self-reported their ethnicity as Hispanic in the study group. For all pregnancy-level data, we limited our sample to pregnancies within 3 years of the interview so that our contraceptive analysis used the same cohort of pregnancies as our intention analysis. Our examination of women who reported ever having had an unintended pregnancy had no exclusions and included any pregnancy at any time period before the interview.

Measures

Outcomes: pregnancy intention, pregnancy outcome, and contraception

Intention was defined as intended, mistimed or unwanted based upon respondents' rating of each pregnancy and the answer to the following two questions: “So would you say you became pregnant too soon, at about the right time, or later than you wanted?” and “Right before you became pregnant did you yourself want to have a(nother) baby at any time in the future?”^7,21 Intended pregnancies were classified as having occurred at the right time or later than desired, mistimed pregnancies were classified as occurring too soon, and unwanted pregnancies occurred when a respondent felt that she did not want another pregnancy at any time in the future. We included answers of “didn't care” in the intended category as per a previous NSFG publication,⁷ and we also included “didn't know” in the intended category since only 1% of our study population provided that response. Unintended pregnancies are defined as both mistimed and unwanted pregnancies. We classified pregnancy outcomes as live birth, elective abortion, current pregnancy, or pregnancy loss, including miscarriage, stillbirth, and ectopic pregnancy.

For contraception, we dichotomized the group of unintended pregnancies within three years of the interview into those preceded by contraception in the month of conception (use) and those preceded by no contraception (non-use). For the use group, respondents were asked to state the methods employed prior to the unintended pregnancy. We categorized these methods into oral contraceptive, patch or ring, condoms, injectable hormones, withdrawal, emergency contraception, sterilization, or long-acting reversible contraception, including intrauterine devices (IUDs) or hormonal implants. All other methods were either classified as other.

Socio-demographic covariates

These variables were chosen based upon their prior association with unintended pregnancy and/or contraceptive use in either the general childbearing population or in the Hispanic population.^6,8,21 Age was categorized per prior NCHS/NSFG reports: 15–19, 20–24, 25–34 and ≥35 years old.^3,7 For the variables income adjusted for household size and educational attainment, we chose cut-points as close to the median as possible for each variable amongst the cohort of women. Results were not sensitive to other cut-points. We categorized self-reported religious belief as Catholic, Protestant, none, or other. Variables relevant to the Hispanic population include immigration status (yes/no), preferred language (Spanish/English), and Hispanic nationality, classified as Mexican or other.

Analyses

From the NSFG pregnancy file and consistent with other NSFG publications,^6,7,22 we determined the proportion of unintended pregnancies among self-reported Hispanic women who reported pregnancies within three years of the NSFG interview.

Pregnancy outcomes were stratified by pregnancy intention to assess relationship between intention and outcome, specifically abortion, and to determine the percentage of unintended pregnancies that ended in abortion. It has been widely documented that abortion is significantly underreported in the NSFG.^6,7,27,28 The 2006–2010 NSFG calculated that only 35% of all abortions that were performed were actually reported to the NSFG.²⁸ Since there is no comparable estimate specifically for the Hispanic population, we used this figure to estimate the additional number of unintended pregnancies ending in abortion that would be underreported by Hispanic respondents.

We determined relative risk for unintended pregnancy for Hispanic women who reported any pregnancy in any time period prior to their interview. We determined the adjusted risk for an unintended pregnancy using the variables found to have a crude association as well as each respondent's total number of pregnancies, another suggested predisposing factor for unintended pregnancy. ⁶ These estimates of relative risk have not been provided in prior publications describing risk factors for unintended pregnancy.^6,7

Finally, we evaluated contraceptive use prior to the unintended pregnancies within 3 years of the NSFG interview. We tested for any difference between the use and non-use group and sociodemographic covariates. We also assessed the reasons respondents from both groups chose as to why they thought their unintended pregnancy occurred dichotomized by use.

Statistics

All results are presented as nationally weighted estimates and proportions based on survey weights obtained from the NSFG. Statistical analysis was performed with Stata Version 12 (College station, TX). To test the significance of bivariate associations, we used Stata complex survey commands for Pearson's Chi-Square test of proportions. In multivariable regression, we used a survey command that produced a modified Poisson regression model with a linearized variance estimator which provides more accurate estimates of risk for the binary outcome of unintended pregnancy.^29,30,31 The model only included variables found to be significant in the bivariate analysis. Since this is a study of an existing, publicly available, de-identified data set, it was not considered human subject research and was not reviewed by the Northwestern University Institutional Review Board

Results

Figure 1 illustrates the steps taken and the populations used for our analysis. Twenty-two percent (n=2723) of the 12,279 women in the female respondent file self-reported their ethnicity as Hispanic. This translates to a weighted population of over 10.4 million Hispanic women, estimated at the 2008 midpoint of the survey time period. This finding is consistent with U.S. Census data from 2010, which showed the number of Hispanic women between the ages of 15 and 44 to be 10.8 million.³² Of note, almost 70% of the overall Hispanic female population cited at least one pregnancy, with over 95% women over the age of 34 and 18% of adolescents, reporting at least one pregnancy in their lifetime (data not shown).

FIG. 1.

Study sample.

From the NSFG pregnancy file, 983 pregnancies to Hispanic women were identified within three years of their interview. The corresponding nationally weighted sample size is approximately 3.8 million pregnancies over three years. While there are no comparable nationally collected pregnancy data, our finding is consistent with the National Center for Vital Statistics birth data, which reported between 900,000 to 1 million births per year to Hispanic women between 2003 and 2010.²

Table 1 presents data on these pregnancies stratified by respondents' self-reported pregnancy intention. We assessed for associations between pregnancy intention and sociodemographic variables as well as pregnancy outcome. Half (51%), or approximately 1.95 million, of Hispanic pregnancies were described as unintended, including 81% of pregnancies among teenagers. In general, unintended pregnancies occurred more frequently among younger, non-immigrant, English speaking, non-religious and single women. There was no significant association between unintended pregnancies and differing levels of educational status, income, or Hispanic group of origin.

Table 1.

Pregnancy Intention and Outcome by Intention for Pregnancies to Hispanic Women^* National Survey of Family Growth, 2006–2010

	% total pregnancies n=983 pregnancies; weighted n=3,804,680	% unintended pregnancies n=481 pregnancies(51%); weighted n=1,946,188
	Weighted proportions
	100	51
Age at conception (years)†
15–19	17	81
20–24	25	54
25–34	45	38
≥35	13	51
Marital status†
Single	27	73
Cohabitating	32	52
Married	41	36
Language†
Spanish	40	44
English	60	56
Immigrant†
Yes	51	45
No	49	58
Educational attainment (.1108)
<High school grad/GED	48	55
≥High school grad/GED	52	47
Income§
<0–149	62	51
≥150	38	51
Hispanic group
Mexican	69	51
Other	31	51
Religion†
Protestant	24	48
No religion	14	66
Other	4	37
Catholic	59	50
Pregnancy outcome†,γ
Live birth	58	53
Elective abortion	8	15
Pregnancy loss	18	18
Current pregnancies at time of interview	16	14

^*Pregnancies within three years of NSFG interview.

^†p<0.001 for unintended vs intended pregnancies.

^§Percent of poverty line, adjusted for household size; calculated using poverty threshold from year.

^γFor pregnancy outcomes, columns, rather than rows, add to 100%.

GED, General Educational Development (high school equivalency) test.

Of the unintended pregnancies, the outcomes were 53% live birth, 15% elective abortion, 18% loss, and 14% current pregnancies. Given that abortion is underreported in the survey,²⁷ our finding of 15%, or 286,090, unintended pregnancies ending in abortion does not reflect the true incidence of abortion in this group. Adjusting our weighted results with the information provided by the NSFG results in an additional half million unintended pregnancies potentially ending in abortion for Hispanic women.²⁷

Table 2 contains our analysis on pregnancy intention from the 1880 Hispanic women in the respondent file who reported having had at least one pregnancy. This translates to a nationally weighted sample size of 7.3 million Hispanic women with a history of at least one pregnancy at the midpoint of the survey time period. Of those women, 69%, or just over 5 million reported at least one of those pregnancies being unintended as compared to 31% reporting never having had an unintended pregnancy. Virtually all (92%) of adolescents who reported a pregnancy, divulged a history of an unintended pregnancy. Bivariate analysis found that younger, single women, U.S.-born women, English speaking, and multigravid women were significantly more likely to have reported at least one unintended pregnancy. Neither a woman's income nor her education or Hispanic group of origin was associated with having had an unintended pregnancy. Our regression analysis found that for the variables with a significant bivariate association, all, with the exception of immigration status and language, remained significant. The strongest risk for unintended pregnancies was found in teenagers, (incidence rate ratio [IRR] 1.64, 95% confidence interval [CI]: 1.44–1.88) and in women with a history of three or more pregnancies (IRR 1.77, 95% CI: 1.53–2.06)

Table 2.

Predictors of an Unintended Pregnancy Among Ever-Pregnant Hispanic American Women National Survey Family Growth, 2006–2010

	% of ever-pregnantHispanic respondents n=1880 women; weighted n=7,307,156	% of ever-pregnant Hispanic respondents reporting n=1307 women (69%); weighted n=5,033,113
	Weighted proportions		Adjusted IRR	95% CI
Age (years)†
15–19	5	93	1.64	(1.44–1.88)
20–24	13	73	1.23	(1.08–1.39)
25–34	40	65	0.99	(0.89–1.11)
>34	42	68	Ref
Marital status‡
Single	27	84	1.34	(1.24–1.46)
Cohabitating	19	70	1.14	(1.00–1.29)
Married	54	61	Ref
Number pregnancies‡
≥3	55	77	1.77	(1.53–2.06)
2	26	61	1.32	(1.12–1.55)
1	19	56	Ref
Language‡
Spanish	45	63	0.93	(0.80–1.08)
English	55	73	Ref
Immigrant†
Yes	46.85	64	0.93	(0.78–1.11)
No	43.18	75	Ref
Educational attainment
Less than high school	43	71
High school diploma or GED	57	67
Household income§
150+	43	66
Less than 150	57	71
Hispanic group
Mexican	68	69
Other	32	68
Religion
Protestant	24	67
No religion	12	80
Other religion	4	75
Catholic	60	67

^†p<0.05 for any unintended versus no unintended pregnancy.

^‡p<0.001 for any unintended versus no unintended pregnancy.

^§Percent of poverty line, adjusted for household size; calculated using poverty threshold from year.

CI, confidence interval; IRR, incidence rate ratio.

From the pregnancy file, 481 of the 983 pregnancies to self-reported Hispanic women within three years of the NSFG interview were identified as unintended. As previously mentioned, this translates to just under 2 million unintended pregnancies. Table 3 shows that among these unintended pregnancies, half were conceived when a woman was not using any contraception and that this behavior did not vary significantly by any of the sociodemographic variables found to be related to unintended pregnancy. Of the pregnancies that were preceded by contraceptive use, the most frequently used methods included oral contraceptives, rings or patch, condoms, or withdrawal. The most common method used prior to unintended pregnancies among teenagers was condoms, which they reported occurring prior to 40% of their pregnancies (data not shown). Finally, Table 4 illustrates the reasons why women felt an unintended pregnancy occurred categorized by contraceptive use during the month of conception. The most common reason reported for an unintended pregnancy despite contraception was improper use (45%), and among the pregnancies preceded by no use, the most common response for not using contraception (37%) was “I did not think I could get pregnant.”

Table 3.

Contraceptive Nonuse Versus Use Prior toSelf-Reported Unintended Pregnancies to Hispanic Women, National Survey of Family Growth Interview, 2006–2010^*,†,‡

	% of total unintended pregnancies n=481 pregnancies; weighted n=1,946,189	% of unintended pregnancies with no contraceptive use prior to conception n=243 pregnancies (50%); n=980,775
	Weighted proportions
Age at conception (years)
15–19	28	49
20–24	26	46
25–34	34	52
≥35	12	61
Educational attainment
Less than high school	52	55
High school+	48	55
Marital status at conception
Single	39	53
Cohabitating	33	43
Married	29	55
Language
Spanish	34	59.63
English	66	45.61
Immigrant
Yes	44	53
No	56	48
Household income§
0–149	62	54
150+	38	44
Hispanic group
Mexican	69	53
Other	31	44
Religion
Protestant	22	51
No religion	19	44
Other religion	3	82
Catholic	57	51

^*Pregnancies within three years of NSFG interview.

^†Contraceptive methods used included: oral contraceptive/ring/patch (22%), condom (21%), withdrawal (16%), not actually using contraception (11%), emergency contraception (4%), intrauterine device or implant (3%), injection (1%), sterilization (0.3%), other (22%, including included calendar, insert, or missing).

^‡No significant differences found.

^§Percent of poverty line, adjusted for household size; calculated using poverty threshold from year prior to interview.

Table 4.

Female Respondents' Self-Reported Reasons for Unintended Pregnancies by Contraceptive Use, National Survey of Family Growth interview, 2006–2010

Reasons why pregnant despite contraception use	% of 243 unweighted, n=980,775 weighted, unintended pregnancies*
Improper use	45
Contraceptive failure	42
Don't know, missing	13

Reasons why no contraceptive use	% of 238 unweighted, n=965,414 weighted, unintended pregnancies*
Did not think could get pregnant	37
Not expecting sex	17
Did not really mind if got pregnant	16
Worried about side effects	12
Male partner himself did not want to use contraception	6
Male partner did not want respondent to use contraception	3
Unable to obtain contraception	1
Don't know/missing	8

^*Pregnancies within three years of NSFG interview.

Discussion

Our findings reveal that while the overwhelming majority of Hispanic women have had a pregnancy, almost 70% of them have had at least one unintended pregnancy and over half of pregnancies to Hispanic women are unintended. To our knowledge, this is the first nationally representative estimate where women are the unit of analysis. However, our finding of 51% unintended pregnancies is within the range of previously published estimates from NSFG data in which pregnancy is the unit of analysis—48% in 1994, 54% in 2001, and 53% in 2006.^6,22 Virtually all (92%) of the almost one in five Hispanic teenagers who have ever been pregnant reported an unintended pregnancy and 81% of all pregnancies to teens were identified as unintended. Given that abortions are underreported, our findings may still be an underestimate of the true frequency of unintended pregnancy in Hispanic women. Regardless of the exact number of women and pregnancies, these are the women whose pregnancies generate the documented disparities in both unintended and adolescent births for the Hispanic population as compared to their non-Hispanic counterparts.

Diminishing this gap requires primary prevention of unintended pregnancies, either via reduced sexual activity or increased effective use of contraception among Hispanic women. Our findings highlight the extent to which Hispanic American women of all ages are not using contraception before an unintended pregnancy occurs. Just under 1 million, or half, of all identified unintended pregnancies to Hispanic women were associated with non-use of contraception prior to conception. Based on our regression model, younger and multigravid women are at higher risk for unintended pregnancy and would likely benefit the most from primary prevention efforts.

To inform and facilitate population specific interventions, we chose to focus solely on the magnitude of risk within the Hispanic population at the level of the individual woman, rather than across populations or at the level of pregnancy. Since we did not aim to compare Hispanic women to other ethnic groups, we did not intend, and cannot conclude, that all of our findings are unique to the Hispanic population. In fact, some of our findings echo previously published data on unintended pregnancy with pregnancy as the unit of analysis. One such example is our result of “I did not think I could get pregnant” being the most common reason for foregoing contraception in our sample of unintended pregnancies.³³ While the complexity of behavior cannot be explained by one survey question, this finding does suggest inadequate reproductive knowledge in the Hispanic population and previous studies have supported this hypothesis when a comparative analysis to other groups was performed.

For instance, two cross sectional analyses found Hispanic women to have less knowledge about contraception when compared to non-Hispanics.^34,35 Garces-Palacio et al. also evaluated reproductive knowledge and found that Hispanic women scored lower than non-Hispanics. Examples of questions where Hispanics more often gave the incorrect answer included: “a woman cannot get pregnant if it is her first time to have sex,” “a woman cannot get pregnant if her menstrual cycles are irregular,” and “if a man pulls out before ejaculation, the woman will not get pregnant.”³⁴

Reasons why Hispanic women have less accurate knowledge about both contraception and reproduction may be reflected in the context, quality, and the quantity of counseling they receive both in the community and in the clinical setting. Hall et al. have published two studies using NSFG data from 2002 and 2006–2010 that showed Hispanic women and women born outside the US used sexual and reproductive health services less frequently than other groups.^36,37 Meneses et al. reported that Hispanic mothers were less likely than white or African American mothers to discuss sex and contraception and the most recent Youth Risk Behavior Surveillance conducted by the U.S. Centers for Disease Control and Prevention found the lowest levels of sex/sexually transmitted disease education among Hispanic students as compared with their non-Hispanic counterparts.^10,38 Recent data also reveal that a Hispanic woman's social network is another influential source of information and potential misinformation about reproduction and contraception.³⁹

When contraception was used prior to the unintended pregnancies in our sample, coital or user-dependent methods—which have the highest known failure rates—were the most common methods cited.⁴⁰ In unintended pregnancies to teenagers, 40% were preceded by condom use. These results underscore why the most common reason identified for an unintended pregnancy despite contraception was “improper use” and why recent publications have emphasized counseling for long acting reversible contraception in women, especially adolescents, at risk for unintended pregnancies.^41,42

While generalizability is the strength of this study, the substantial limitation of recall bias and, specifically for this study, the underreporting of unintended pregnancies must also be acknowledged. Our abortion data suggest substantial underreporting of unintended pregnancies. The survey questions themselves may also lead to bias. Several publications have suggested that labeling pregnancy intention as a categorical phenomenon oversimplifies an extremely sensitive and complex experience.^21,43,44 Additionally, women may be reticent to retrospectively report a pregnancy as unintended when it resulted in the live birth of one or more of her children.

Limitations also surfaced with respect to the finite number and presentation of variables within the NSFG dataset. For example, due to the structure of the interview, we could not calculate insurance status and/or access at the moment of conception or receipt of reproductive health services relative to the timing of the conception of an unintended pregnancy. The NSFG also lacks variables related to history of violence, mental illness, or substance abuse that are also potential correlates of unintended pregnancy.^45–49

Finally, unlike previous publications, we repeatedly found no association, even in bivariate analysis, between education or income and unintended pregnancy or contraceptive use. These results did not differ by cut-point modification and were consistent when both pregnancies and women were used as the denominator. We also found no association in multivariable regression between immigration status and language, common proxies for acculturation, and pregnancy intention and contraceptive use. Both of these unexpected findings suggest that intervention programs for this population may not need to distinguish women by level of acculturation and that all Hispanic women, especially adolescent and multigravidas, should be included in primary prevention initiatives for unintended pregnancy.

Conclusions

Our results highlight the broken link between pregnancy intention and contraceptive use in the Hispanic population. There remains a continued need to better educate and empower Hispanic women and girls about their reproductive capacity and their contraceptive practices. Further research should strive to more clearly elucidate unintended pregnancy primary prevention approaches and inform future family planning strategies for the Hispanic population.

Acknowledgments

This study was supported by an institutional award for postdoctoral training to the Northwestern University Feinberg School of Medicine Center for Healthcare Studies from the Agency for Healthcare Research and Quality, T-32 HS 000078. It was also supported by grant number P01HS021141 from the Agency for Healthcare Research and Quality.

Disclosure Statement

The authors have no conflict of interest and no competing financial interests exist.