SHLAY, JUDITH C. MD, MSPH*†‡; MAYHUGH, BARBARA RN, OB/GYN NP*; FOSTER, MARK BS, MA*; MARAVI, MOISES E. BS, MS*
UNINTENDED PREGNANCY is an important and complex problem that has significant public health consequences. Previous studies indicated that 49% of pregnancies in the United States in 1994 were unintended, and 54% ended in therapeutic abortions. 1 Almost all women are at risk for unintended pregnancy throughout their reproductive years. However, rates of unintended pregnancies are higher (≥75%) in certain populations, such as adolescents, never-married women, women with low incomes, and those not using contraception. 1,2
Sexually transmitted disease (STD) and unintended pregnancy are related in a number of important ways. 3 Both require sexual contact, usually intercourse, and in both it is usually women who suffer the major consequences. STDs are more easily transmitted to women and are more difficult to diagnose since they are more frequently asymptomatic in women than in men, which often leads to serious sequelae. The biologic burden of unintended pregnancy falls entirely on women. Furthermore, the demographic characteristics of groups of women most affected by STD and unintended pregnancy is similar: poor, minority women younger than 25 years of age have the highest rates of both STD and unintended pregnancies. 4 Although STD clinics encourage the use of barrier methods of contraception to prevent the spread of STD, generally limited attention is given to non-STD reproductive healthcare needs. 5
Previous research conducted mainly among adolescents has identified certain social contextual influences that increase the risk for pregnancy. 6,7 Scaramella et al 6 determined that risk-taking behavior in later adolescence mediated by the lack of parental involvement, poor academic competence, and deviant-peer affiliations increased the risk of pregnancy. Identifying determinants of pregnancy in women attending an STD clinic can help to target preventive efforts. The purpose of this study was to identify predictors for women attending an STD clinic who are at risk for pregnancy.
This study utilized a subcohort of women enrolled in a randomized, controlled, nonblinded trial between December 1996 and December 1999 that evaluated the benefit of clinic-initiated contraceptive services on the implementation of consistent ongoing contraceptive care and prevention of unintended pregnancy and STD.
The study was conducted at the Denver Metro Health Clinic (DMHC), the STD clinic operated by Denver Public Health, a department of Denver Health (DH). DH is a comprehensive public hospital system that includes an acute care hospital, a community health clinic program, and public health clinics. The DMHC is a readily accessible and confidential facility offering free comprehensive, walk-in services for diagnosis, treatment, management, and prevention of STD. Services are provided to all interested persons, regardless of county of residency, in 15,000 to 18,000 visits per year.
Women presenting to the DMHC with a new problem who identified themselves as being residents of the Denver metropolitan area were asked at registration to complete a brief screening survey about current contraceptive practices in order to determine eligibility for the study. Only premenopausal women who were not using any contraception or who were using only a method considered to be suboptimal (e.g., condoms, withdrawal, natural family planning), were not currently pregnant or wanting to become pregnant, and were not planning to move from the metropolitan area were asked to participate. Any woman who chose to enroll completed an informed consent form and underwent pregnancy testing, of which the results had to be negative, before randomization.
Enrolled individuals completed a routine STD evaluation and a baseline questionnaire detailing demographic information, sexual history, reproductive history, previous birth control use, previous birth control side effects, condom use, and attitudes about contraception and condom use as well as pregnancy prevention.
Subjects were randomly assigned to either the intervention group or the control group. Women in the intervention group were counseled by a family planning nurse about contraceptive options and had the method of their choice initiated in the clinic, at either the enrollment or an early follow-up visit, following which they were assisted with making an appointment with a primary care provider (PCP) of their choice for ongoing contraceptive care. Women in the control group were provided with brief advice about contraceptive options and standard information about clinics and providers that they might contact for contraceptive services. All participants were provided with condoms and spermicide at their initial STD clinic visit.
Women enrolled in the project were asked to consent to a brief interview at 4, 8, and 12 months after enrollment, a review of their medical records at the office or clinic of any subsequent PCP for assessment of various study outcomes, a review of three statewide databases of reported STD for STD outcome, and the state's computerized Birth Registry for pregnancy outcomes. These follow-up interviews included information about birth control methods used over the past 12 months, name of any PCPs seen for contraceptive care during the prior 12 months and date first seen, interval pregnancy, and STD history. The institutional review board approved the study.
Data were analyzed with SAS version 8.2 software (SAS Institute, Cary, NC). Based on initial frequencies, five age categories were created: (1) <20 years, (2) 20 to 24 years, (3) 25 to 29 years, (4) 30 to 39 years, and (5) 40 to 49 years. Association of baseline demographic, behavioral, and clinical variables with incident pregnancy was assessed by chi-square analysis. Logistic regression modeling was performed with use of variables found to be significant by univariate analysis at P values ≤0.20 or ones that were considered clinically relevant.
The study population comprised 877 women, whose demographic and clinical characteristics are outlined in Table 1. The majority of patients were ≤24 years old (40.1%, ≤19 years; 35.8%, 20–24 years; and 24.1%, >25 years) and of nonwhite ethnicity (whites, 39.5%; blacks, 26.1%; Hispanics, 30.0%; and other, 4.4%). Most women had never married (85%), and the majority of women (74%) had obtained (at the most) a high school/general equivalency diploma (GED); only 26% had had at least 1 year of college.
Of the study participants, 55% had experienced at least one previous pregnancy, and 31% reported multiple previous pregnancies. Forty-eight percent of those previously pregnant had had a therapeutic abortion. The median age at the first pregnancy was 17 years (range, 9–36), and the average number of previous pregnancies was 2.4 ± 1.9. Only 5.4% had used any form of effective contraception (e.g., oral contraceptives, Depo-Provera[Pharmacia & Upjohn], vasectomy, intrauterine device, or Norplant [Wyeth-Ayerst Laboratories]) over 75% of the time during the past 4 months; the remainder had used less-effective methods (e.g., condoms, spermicide, withdrawal, natural family planning) or no method at all. Only 27% reported currently having a PCP, and the main type of provider was a private physician (39%) or a community health clinic (32%). The majority of participants did not have health insurance (61%).
At enrollment, 57% described a past STD infection, and 41% had an STD diagnosed at that visit to the STD clinic. Twenty-two percent of the enrolled participants had had between 3 and 5 sex partners during the past month; 40% reported >75% condom use over the past 4 months. Twenty percent believed that they had or may have been exposed to HIV over the past year.
Comparing the baseline characteristics of the entire study population (n = 877) with those of women for whom follow-up pregnancy information was available (i.e., occurrence of pregnancy or no pregnancy was confirmed; n = 673), only having a history of STD was significantly different between the two groups (P = 0.03), occurring more often in the subgroup. The remaining baseline characteristics were not statistically different between the two groups.
Of the enrolled participants, pregnancy information was available for 673 (76.7%), and 220 incident pregnancies occurred within 1 year after enrollment (32.7%). By univariate analysis, an incident pregnancy was found to be associated with age ≤19 years (OR, 2.4; 95% CI: 1.6–3.7), being nonwhite (OR, 1.8; 95% CI: 1.3–2.6), having at the most a high school diploma or a general equivalency diploma (OR, 1.8; 95% CI: 1.2–2.7), having had at least one previous pregnancy (OR, 1.6; 95% CI: 1.2–2.3), having had a previous therapeutic abortion (OR, 2.8; 95% CI: 2.0–3.9), age of <17 years at first pregnancy (OR, 1.9; 95% CI: 1.4–2.7), having not used any birth control method in the most recent sexual encounter (OR, 1.6; 95% CI: 1.2–2.3), having had sex at least once a week (OR, 1.7; 95% CI: 1.2–2.4), and having either a chlamydial infection (OR, 2.3; 95% CI: 1.5–3.7) or a gonorrhea infection (OR, 1.9; 95% CI: 1.1–3.5) at the time of enrollment (Table 2).
By sequential logistic regression analysis controlling for study assignment, incident pregnancy was associated with age ≤19 years (OR, 2.8; 95% CI: 1.5–5.2), previous abortion (OR, 3.1; 95% CI: 1.7–5.4), frequency of sexual encounters of at least once a week (OR, 1.8; 95% CI: 1.2–2.6), and having a chlamydial infection at the time of enrollment (OR, 1.8; 95% CI: 1.0–3.2). Overall, there was no independent contribution in the prediction of incident pregnancy with racial/ethnic groups, pregnancy history, educational level, nonuse of birth control during most recent intercourse, age at first pregnancy, number of sex partners within the past month, or current gonorrhea. When only variables amenable to behavior change were included in the model, incident pregnancy was associated only with frequency of sexual encounters of at least once per week (OR, 1.7; 95% CI: 1.1–2.5) and having a chlamydial infection at the time of enrollment (OR, 2.1; 95% CI: 1.2–3.7).
By combining the demographic and behavioral characteristics, which were correlated with incident pregnancy by univariate analysis, in a sequential manner, we found that the cumulative risk of pregnancy for women with 6 or more of the 9 characteristics was 51%, compared with 25.6% for women with ≤5 characteristics. Of this group, 27% had ≥6 of the 9 characteristics that were included in this analysis. The baseline characteristics placing women at higher risk for an unwanted pregnancy included age of ≤19 years, being nonwhite, having at the most a high school diploma or a GED, having had at least one pregnancy, having not used any birth control method in the last sexual encounter, having had sex at least once a week for the past 4 months, having had an abortion, having ≥3 sex partners within the past month, and being <17 years of age at the time of the first pregnancy.
This study evaluated the relationship between certain demographic and behavioral characteristics and the probability of an incident pregnancy in women attending an urban STD clinic. We found that by using multivariate analysis and controlling for study assignment, incident pregnancy was predicted only in women presenting to the STD clinic who were younger, had had at least one abortion, were currently infected with Chlamydia, and had had sexual intercourse at least once a week over the past 4 months, before enrollment in the study (Table 2). However, in assessment of the cumulative risk using variables that were correlated with incident pregnancy, the model revealed that demographic and behavioral characteristics could be combined to identify a subgroup whose pregnancy rate was 51%, compared with 25.6% for those women with five or fewer characteristics. Thus, although a particular risk variable may not be able to exert a substantial independent effect in multivariate models that include several risk predictors, it may affect outcomes as a consequence of the conjoint influences of all of the risk factors put together. The cumulation of risk, many argue, is the greatest predictor of both positive and negative outcomes. 8
Previous research on the factors associated with pregnancy prediction has involved primarily adolescents. 6,7,9–11 Predisposing factors have included a number of characteristics found to be associated with pregnancy (e.g., housing situation, history of abuse, deviant behavior, psychological profile) that would be impossible to assess in the course of an STD clinical evaluation. Furthermore, many of the studies have illustrated the complexity of human behavior and the inherent difficulty in predicting incident pregnancy. 6,7
Goldfarb et al, 11 using cumulative risk theories, have examined both the sociodemographic risk and the protective factors associated with adolescent pregnancy. By using an inferential strategy to detect “pregnancy susceptibility” in an indigent population, they were able to predict pregnancy. 11 The profile developed indicated that if an indigent adolescent came from a large family, received sex education late and from an acquaintance, and had poor academic performance, she was then more likely to have an unplanned pregnancy, in comparison with an indigent adolescent who came from a smaller family, attained normal grade level, and received her initial sex education early and from an adult in the home. This strategy showed that pregnancy was a consequence of the multiple influences that, when looked at cumulatively, were predictive of higher risk of a subsequent pregnancy. However, the limitation of this strategy and that developed by Kalil et al 10 was that these approaches were evaluated only for adolescents and were not assessed for adults, for whom there would be different factors concerning their current living situation. Furthermore, neither of these approaches would be practical to incorporate into an STD clinical evaluation.
Since the women participating in this study indicated at enrollment that they had no intention of becoming pregnant, a requirement for participation in the study, we considered all incident pregnancies as being unintended. Given that intention has been linked to important pregnancy-related behaviors (e.g., recognizing the signs of pregnancy earlier, seeking out prenatal care earlier, and quitting smoking), predicting unintended pregnancy in high-risk women is important. 12 By using a combination of demographic and behavioral characteristics that were identified in this study and are readily available through a clinical examination, it is possible to identify a subgroup at much higher risk for subsequent pregnancy and thus initiate counseling about pregnancy avoidance.
Our investigation has several potential limitations. First, since this study was conducted in an STD clinic, the results may not be generalizable to other clinical populations. Second, because of limitations in the questionnaires used in the STD clinic, certain sociodemographic characteristics (e.g., age at first intercourse, housing situation) that previously have been shown to be useful in predicting incident pregnancy were unknown. Third, because some participants did not return for the follow-up interviews, no pregnancy information was available for approximately 25% of women enrolled in the study.
Finally, vital statistics reflect only delivery information, not pregnancy intention. Because some women were lost to follow-up, the pregnancy rate was calculated from the follow-up interviews and by using vital records to supplement self-reports. Thus, incident pregnancy was used as the outcome instead of unintended pregnancy, which potentially could influence the results, although all of the women indicated at enrollment that they were not intending to become pregnant.
In conclusion, our study showed that by assessing cumulative risk, a combination of demographic and behavioral characteristics were useful in identifying a subgroup of women at higher risk for subsequent pregnancy. Thus, by using a risk assessment of particular characteristics in combination that would be easy to assess in a clinical setting, STD clinical providers would be able to target an intervention for this high-risk group through the provision of counseling for both pregnancy and STD prevention.
1. Henshaw SK. Unintended pregnancy in the United States. Fam Plann Perspect 1998; 30: 24–29.
2. Forest JD. Epidemiology of unintended pregnancy and contraceptive use. Am J Obstet Gynecol 1994; 170: 1485–1488.
3. Cates W Jr. Sexually transmitted diseases and family planning: strange or natural bedfellows, revisited. Sex Transm Dis 1993; 20: 174–178.
4. Upchurch DM, Farmer MY, Glasser D, Hook EW III. Needs and practices among women attending an inner-city STD clinic. Am J Public Health 1987; 77: 1427–1430.
5. Horn JE, McQuillan GM, Ray PA, Hook EW III. Reproductive health practices in women attending an inner-city STD clinic. Sex Transm Dis 1990; 17: 133–137.
6. Scaramella LV, Conger RD, Simons RL, Whitbeck LB. Predicting risk for pregnancy by late adolescence: a social contextual perspective. Dev Psychol 1998; 34: 1233–1245.
7. Stiffman AR, Earls F, Robins LN, Jung KG, Kulbok P. Adolescent sexual activity and pregnancy: socioenvironmental problems, physical health and mental health. J Youth Adolesc 1987; 16: 497–509.
8. Sameroff A, Bartko WT, Baldwin A, Baldwin C, Siefer R. Family and social influences on the development of child competence. In: Lewis M, Feiring C, eds. Families, Risk and Competence. Mahwah, NJ: Lawrence Erlbaum, 1998: 161–185.
9. Vernon MEL, Seymore C. Antecedents and predictors of adolescent pregnancy. Adv Adolesc Ment Health 1990; 4: 109–119.
10. Kalil A, Kunz J. First births among unmarried adolescent girls: risk and protective factors. Soc Work Res 1999; 23: 197–208.
11. Goldfarb JL, Mumford DM, Shurm DA, Smith PB, Flowers C, Shum C. Psychosocial factors associated with pregnancy in adolescent girls: a preliminary report. J Nerv Ment Dis 1977; 6: 127–144.
12. Kost K, Landry DJ, Darroch JE. The effects of pregnancy planning status on birth outcomes and infant care. Fam Plann Perspect 1998; 30: 223–230.