This article was downloaded by: [Lancaster University Library] On: 21 October 2014, At: 03:16 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Women & Health Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/wwah20
Partner Violence and Abortion Characteristics a
Lisa Colarossi PhD LCSW & Gillian Dean MD MPH
a
a
Planned Parenthood of New York City, New York, New York, USA Accepted author version posted online: 28 Feb 2014.Published online: 08 Apr 2014.
To cite this article: Lisa Colarossi PhD LCSW & Gillian Dean MD MPH (2014) Partner Violence and Abortion Characteristics, Women & Health, 54:3, 177-193, DOI: 10.1080/03630242.2014.883662 To link to this article: http://dx.doi.org/10.1080/03630242.2014.883662
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Women & Health, 54:177–193, 2014 Copyright © Taylor & Francis Group, LLC ISSN: 0363-0242 print/1541-0331 online DOI: 10.1080/03630242.2014.883662
Partner Violence and Abortion Characteristics LISA COLAROSSI, PhD, LCSW and GILLIAN DEAN, MD, MPH
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Planned Parenthood of New York City, New York, New York, USA
We conducted a retrospective cohort study using randomly selected medical charts of women reporting a history of partner violence and women with no history of partner violence at the time of a family planning or abortion appointment ( n = 6,564 per group). We analyzed lifetime history of partner violence for odds of lifetime history of abortion and miscarriage number, and birth control problems. To more closely match timing, we analyzed a subsample of 2,186 women reporting current violence versus not at the time of an abortion appointment for differences in gestational age, medical versus surgical method choice, and return for follow-up visit. After adjusting for years at risk and demographic characteristics, women with a past history of partner violence were not more likely to have ever had one abortion, but they were more likely to have had problems with birth control, repeat abortions, and miscarriages than women with no history of violence. Women with current partner violence were also more likely to be receiving an abortion at a later gestational age. We found no differences between the groups in return for abortion follow-up visit or choice of surgical versus medication abortion. Findings support screening for the influence of partner violence on reproductive health and related safety planning. KEYWORDS control
abortion, partner violence, miscarriage, birth
INTRODUCTION More than two decades of research shows associations between partner violence and detrimental sexual and reproductive health outcomes including Received June 5, 2013; revised January 3, 2014; accepted January 9, 2014. Address correspondence to Lisa Colarossi, PhD, LCSW, Planned Parenthood of New York City, 26 Bleecker Street, New York, NY 10012. E-mail: [email protected] 177
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
178
L. Colarossi and G. Dean
lower use of condoms and other contraceptives; higher rates of sexually transmitted infections (STIs), HIV, urinary tract infections, unintended and teen pregnancies; poor birth outcomes; and miscarriage due to physical assault (Coker, 2007; Gazmararian et al., 2000; Plichta & Abraham, 1996; Silverman et al., 2006). In particular, pregnancy is associated with first incidences of partner violence, as well as increased severity (Janeski, 2004; Sagrestano et al., 2004); and pregnancy termination is more likely for women experiencing partner conflicts and violence (Coker, 2007; Ely & Otis, 2011; Foster & Kimport, 2013). However, little is known about whether partner violence is related to abortion characteristics such as method, timing, number, and follow-up. Abortion characteristics have mostly been studied for their association with demographic factors, such as age, race, income, education, marital status, and number of children (Jones & Finer, 2012; Pazol et al., 2011). Similar demographic characteristics are associated with partner violence as well (Rennison & Welchans, 2003; Tjaden & Theonnes, 2000), which may have unique influences on abortion characteristics. New theoretical models propose a number of psychological, social, and physical pathways by which partner violence affects abortion and other reproductive health outcomes (Heise, 1998; Moore, Frohwirth, & Miller, 2010). One important pathway is through reproductive control by a partner, defined as interference with a woman’s ability to make independent/autonomous decisions about her reproduction. Reproductive control can be inflicted economically, emotionally, and/or physically. Characteristics include partner enforcement of his sexual and reproductive intentions in direct conflict with or without consideration of the woman’s intentions via coercion, intimidation, threats, and/or physical assault (Moore et al., 2010). Recent research has provided evidence for the impact of partner refusal to withdraw during sex and/or to use a condom, sabotaging birth control methods, and sexual assault on reproductive outcomes from STIs and HIV transmission to unintended pregnancy and abortion (Ely & Otis, 2011; Moore et al., 2010; Miller et al., 2010; ODonnell et al., 2009; Saftlas et al., 2010; Silverman et al., 2011; Thiel de Bocanegra et al., 2010; Williams, Larsen, & McCloskey, 2008). The few studies that have analyzed partner violence as a variable in abortion research have estimated the proportion of partner violence among women seeking abortion from 11% to 40% (Chavkin & Gee, 2008; Saftlas et al., 2010; Woo, Fine, & Goetzl, 2005). Women experiencing partner violence and fear of negotiating pregnancy decisions with their partner are more likely to have had an abortion and at a later gestational age (Finer et al., 2006; Foster & Kimport, 2013; Jones & Finer, 2012; Jones, Moore, & Frowirth, 2010; Williams & Brackley, 2009), as well as repeat abortions (Ely & Otis, 2011; Fisher & Singh, 2005; ODonnell et al., 2009). Men perpetrating partner violence are more likely to be involved in partner conflicts about whether or not to terminate a pregnancy and to report repeat abortions among their
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Partner Violence and Abortion Characteristics
179
partners (Silverman et al., 2010). Many of these studies have had small sample sizes and have not examined various abortion characteristics within the same study. Our study aim was to compare a large sample of women reporting past and current partner violence to women who did not for differences in a variety of abortion characteristics, after adjusting for some demographic characteristics, years of sexual activity, parity, and problems with birth control. We conducted a retrospective chart review of women receiving abortion and family planning services at three large health centers in New York City. First, we compared women reporting any lifetime history of partner violence to those with no history for differences in lifetime pregnancy outcomes of live births and elective and spontaneous terminations. Second, we analyzed a sub-sample of women obtaining current abortions to compare women reporting current partner violence to those with no current violence for differences in gestational age, medical versus surgical method choice, and return for follow-up visit. In this way we were able to analyze models with closer time-ordering, with lifetime versus current experiences of partner violence and abortion.
METHOD Design and Sampling We conducted a retrospective cohort study using data from medical charts of women receiving a family planning or abortion service over a one-year period (June 1, 2010 to May 31, 2011) at three reproductive health care centers in the Bronx, Brooklyn, and Manhattan in New York City. Family planning services included comprehensive gynecological exams, visits for contraception, pregnancy testing, and STI and HIV testing and treatment. Abortion visits included medical and surgical abortions between 4 and 24 weeks gestational age. From this one-year sampling frame, we randomly selected equal numbers of charts for women (1) reporting any history of partner violence (current or past; N = 6,564) and (2) reporting no partner violence (N = 6,564). An individual woman may have had more than one visit during 2010–2011, and therefore we selected the sample by a singular patient identification and appointment date through computer randomization of electronic medical record data. Information technology staff downloaded de-identified data from electronic medical records into a Micosoft Excel format, which researchers then converted to an IBM SPSS version 20 database for statistical analysis. We received an Institutional Review Board exemption from Mt. Sinai School of Medicine, New York, NY for the use of de-identified medical data in research. Lifetime history of partner violence, abortion number, and miscarriage number were self-reported on the medical history forms of women receiving
180
L. Colarossi and G. Dean
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
reproductive health services. To narrow the temporality of associations among current partner violence and current abortion outcomes, we conducted a sub-analysis of women currently receiving an abortion service from 4 to 24 weeks gestational age from the sample above (N = 5,609; i.e., removing the family planning visits). For these visits, health center staff recorded gestational age and return for post-abortion follow-up visit, which we analyzed for relative risk between women who self-reported current partner violence, past partner violence (but no current), and no partner violence ever. Finally, for women receiving a first trimester abortion (N = 4,958), we analyzed differences in choice of medication versus surgical termination by partner violence group.
Measures DEMOGRAPHIC
CHARACTERISTICS
Demographic characteristics of age, age of sexual debut, race/ethnicity, and payment method were derived from patients’ self-reports. Race/ethnicity was coded into four categories: Black (non-Hispanic), Hispanic, Other, and White (non-Hispanic). We coded payment method into three categories: Medicaid, self-pay, and commercial/private insurance. We analyzed age as a continuous variable in years. We created the variable for number of years at sexual risk by subtracting age of sexual debut from current age and analyzed as a continuous variable in years. We did not analyze other typical demographic characteristics because the electronic medical record had a large amount of missing data for employment status, marital status, and education as many patients do not to report this information and are not required to do so to receive services. PARTNER
VIOLENCE
Partner violence was derived from patients’ self-reported answers to empirically based questions about partner violence in the past year and ever (Colarossi, Breitbart, and Betancourt, 2010a; Zeitler et al., 2006). Three questions asked about partner violence in the past year that included being: (1) hit, kicked, or slapped; (2) threatened or frightened; and (3) forced to have sex. Two questions asked if prior to the last year, any partner had ever (1) hit, kicked, or slapped them and (2) sexually assaulted them. Patients answered these questions in writing on the medical history form, and were asked again verbally by health center staff. We created a dichotomous yes/no variable for any partner violence history by coding yes if an affirmative answer was given to any of the six questions (past and/or current violence) or not. We also created a variable for current partner violence by coding the data into three categories: current partner violence experienced in the past
Partner Violence and Abortion Characteristics
181
year, past partner violence but not in the last year, and no partner violence history. BIRTH
CONTROL PROBLEMS
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Patients reported birth control problems on the medical history form in response to the question, Have you ever had problems with your birth control method? We dichotomously coded the answers into yes/no. No detailed information about the type of problem was recorded on the medical history form, but rather was discussed in counseling with a health care provider. Thus, additional data were not available from electronic medical records. HISTORY
OF PREGNANCY OUTCOMES
History of pregnancy outcomes was derived from self-reported information on the medical history form. We coded number of lifetime miscarriages (spontaneous terminations) and abortions (elective terminations) into three categories: 0, 1, and ≥2 for each outcome. We used number of live births was as a continuous covariate in the prediction of miscarriage and abortion outcomes, which was a measure of parity and not of how many children a woman was currently parenting. ABORTION
CHARACTERISTICS
Health center staff recorded abortion characteristics for women receiving an abortion service on the selected appointment date in in the medical chart. We coded gestational age (via pre-procedure ultrasound) into three categories: ≤12 weeks, 13–18 weeks, and 19–24 weeks, and patient return for a post-abortion follow-up visit as yes/no. We dichotomously coded abortion method as medical or surgical for those women up to 9 weeks gestation who were allowed to choose their termination method after being counseling on their options.
Data Analysis First, we examined bivariate associations between partner violence categories (none any, past only, and current) and demographic characteristics (age, race, and payment method), years of sexual activity, birth control problems, pregnancy outcomes (number of live births, miscarriages, and abortions), and abortion characteristics (gestational timing, method, and return for follow-up). Second, we analyzed separate ordinal logistic regression models for the odds of lifetime number of abortions and miscarriages (both categorized as none, one, or ≥2) by any partner violence history
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
182
L. Colarossi and G. Dean
(yes/no), adjusting for age, number of years of sexual activity, parity, race, payment method, and birth control problems. Each demographic characteristic (age, race, years sex, parity, payment, and birth control problems) was significantly associated with partner violence; therefore, all were included in the regression model to remove the variance associated with the dependent variables before estimating the effects of partner violence (i.e., to adjust for the covariance with partner violence). All covariates remained significant related in the multivariate models with no significant interaction effects of IPV with other covariates; so, we tested only main effects of covariates in the final models. Third, we analyzed separate ordinal and binary logistic regression models for the subsample of abortion patients only (i.e., removing cases that were currently receiving a family planning service) for the odds of gestational timing (1–12, 13–18, and 19–24 weeks) and return for follow-up (yes/no) by current partner violence after adjusting for age, number of years sexually active, parity, race, and payment method. Finally, we analyzed a binary logistic regression model for those cases receiving first trimester abortion services for the odds of currently choosing a medical versus a surgical abortion procedure by current partner violence after adjusting for age, race, and payment method (only first trimester pregnancies have the option of a procedure type). We removed history of birth control problems from all final models related to current abortion characteristics because it was not a significant predictor of these variables, nor did we theoretically expect that it would be a predictor of gestational age, termination method, or return for follow-up. We found no significant interaction effects of IPV with the covariates on the dependent variables, so we tested main effects only in the final models reported here. We found that all models had good fit were by virtue of statistically significant -2 log-likelihood statistics showing that the model variables outperformed the null hypothesis, as well as small, non-significance Pearson and Deviance goodness-of-fit statistics. We used SPSS version 20 for data analysis.
RESULTS Demographics and Partner Violence The sample was highly diverse with ages ranging from 14–60 years (Median = 26.4 years), years of sexual activity ranging from 0–45 years (Median = 10 years), and parity ranging from 0–9 (Median = 0). 35% of women identified as black, 29% as Hispanic (White and non-White), 23% white, and 13% other races. Fifty-nine percent of women paid for their current visit with Medicaid, 36% self-paid, and 5% with commercial/private insurance. Of women reporting any partner violence history, 33% reported violence in the past year.
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Partner Violence and Abortion Characteristics
183
Women reporting any partner violence were slightly older (Mean = 27.8 years, Standard Deviation = 6.6 years) than those reporting none (Mean = 26.95 years, Standard Deviation = 5.96 years), as were those reporting past violence only (Mean = 28.1 years, Standard Deviation = 6.43 years; Table 1). Women reporting current violence were the same age as those reporting none (Mean = 27.0 years, Standard Deviation = 6.7 years). This pattern was similar for years of sexual activity and parity, also with statistically significant but very small differences. Black women had a slightly higher odds of reporting any partner violence versus none compared to White women (odd ratio [OR] = 1.13, confidence interval [CI ] = 1.03–1.23), while Hispanic and others were similar to White women. No racial differences was observed in past violence versus no violence, but Black (OR = 1.53, CI = 1.34–1.76), Hispanic (OR = 1.40, CI = 1.21–1.61) and other women (OR = 1.26, CI = 1.06–1.49) had higher odds of reporting current violence compared to White women. Women reporting any (OR = 1.16, CI = 1.0–1.36) and current partner violence (OR = 1.30, CI = 1.34–1.63) were slightly more likely to be paying with Medicaid than commercial insurance. We found no differences between self-pay and commercial insurance by partner violence experience. In the full sample, 18% of women reported ever having problems using birth control; 9% had a lifetime history of one miscarriage, and 3% had ≥2 miscarriages. A total of 41% of women had never had a therapeutic abortion; 28% had one prior abortion, and 31% had ≥2 abortions. Compared to women with no partner violence, women reporting any history of violence, past and current, were all more likely to have experienced problems with birth control and to have had ≥2 miscarriages. Women who reported current partner violence were more likely to have had one prior abortion (OR = 1.70, CI = 1.44–1.99) than those reporting no violence. We found no differences between women who reported past violence and those who reported no violence for having had one prior abortion. However, those reporting past (OR = 1.24; CI = 1.13–1.36) and current violence (OR = 2.38, CI = 1.79–3.16) were both more likely to have had ≥2 abortions in their lifetime compared to no abortions than women who did not report histories of partner violence. For the subsample of women receiving an abortion on the randomly selected appointment date, those reporting current violence (OR = 1.63; CI = 1.15–2.30) were more likely to have an abortion appointment at a gestational age from 19–24 weeks compared to ≤12 weeks than women with no violence. We found no other differences between partner violence categories and gestational age. We also found no differences between partner violence categories and return for a follow-up visit. For those women who were receiving a first trimester abortion, 70% chose a surgical procedure. We found no association between partner violence and choice to have a medical versus a surgical termination procedure.
184 27.8 (1824) 43.8 (2876)
0 (ref)
28.4 (1864)
81.2 (5327)
>=2
0 (ref) Number of abortions 1
11.4 (751)
7.4 (486)
Number of miscarriages 1
>=2
5.5 (362) 40.3 (978)
37.5 (2464)
Commercial (ref) Birth control problems
Self-pay
57 (3738)
22.1 (1529)
13 (965)
Other
Non-Hispanic White (ref) Payment method Medicaid
29 (1864)
35.9 (2206)
Hispanic
Race/ethnicity Non-Hispanic Black
Characteristic
No partner violence N = 6,564 Column % (n) (ref)
27.1 1.08 34.2 1.39 38.7
11.4 1.65 1.8 2.27 86.8
60.5 1.16 34.5 1.01 5 59.7 1.61
33.6 1.13 28.4 1.08 14.7 .93 23.3
(1, 791) (1.0 − 1.19)p= .◦6 (2, 256) (1.28 − 1.15)∗∗∗ (2, 251)
(751) (1.46 − 1.85)∗∗∗ (117) (1.81 − 2.83)∗∗∗ (5, 696)
(3, 988) (1.0 − 1.36)∗ (2, 278) (.86 − 1.18) (332) (1, 450) (1.47 − 1.76)∗∗∗
(2368) (1.03 − 1.23)∗∗ (1916) (.98 − 1.19) (857) (.83 − 1.05) (1457)
Any partner violence N = 6,564 Column % (n) OR (CI)
(2, 550) (.92 − 1.30) (1, 619) (.72 − 1.15) (225) (989) (1.50 − 1.83)∗∗∗
(1509) (.88 − 1.08) (1260) (.87 − 1.07) (552) (.72 − .93)∗∗ (1073)
24.2 .88 33.4 1.24 42.4
(1, 065) (.80 − .97)∗∗ (1, 464) (1.13 − 1.36)∗∗∗ (1, 865)
11.3 (495) 1.63 (1.43 − 1.83) 3.7 (163) 2.23 (1.75 − 2.84)∗∗∗ 85(3, 736)
58.0 1.10 36.8 .91 5.2 40.7 1.66
34.3 .97 28.7 .96 12.6 .81 24.4
Past N = 4,394 Column % (n) OR (CI)
33 1.64 36 1.84 31
11.7 1.70 3.9 2.38 84.4
65.2 1.30 30 .91 4.8 18.8 1.51
38.8 1.53 29.8 1.40 13.9 1.26 17.5
(721) (1.46 − 1.85)∗∗∗ (788) (1.63 − 2.07)∗∗∗ (677)
(255) (1.44 − 1.99) (86) (1.79 − 3.16)∗∗∗ (1, 919)
(1, 425) (1.34 − 1.63)∗ (655) (.72 − 1.15) (106) (457) (1.33 − 1.71)∗∗∗
(849) (1.34 − 1.76)∗∗∗ (652) (1.21 − 1.61)∗∗∗ (303) (1.06 − 1.49)∗∗ (382)
Current N = 2,186 Column % (n) OR (CI)
TABLE 1 Percent Distribution of Demographic Characteristics and Pregnancy Outcomes by Partner Violence with Unadjusted Odds Ratios (OR) and 95% Confidence Intervals (CI)
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
185
6.9 (189) 89.9 (2, 459) 49 (1, 144)
Gestational age 19–24 weeks
13–18 weeks
1–12 weeks (ref) Follow-up (yes)
10.20 (5.79)
Age (years)
Number of years of sexual activity
27.76 1.02 11.78 1.04 .67 1.12
3.9 1.25 6.9 1.00 89.2 51 1.01 49 1.11
(6.56) (1.01 − 1.03)∗∗∗ (6.40)∗ (1.04 − 1.05)∗∗∗ (1.04)∗ (1.09 − 1.16)∗∗∗
Any IPV N = 6,564 Mean (SD) OR (CI)
(109) (.93–1.66) (192) (.81 − 1.23) (2, 499) (1, 180) (.91 − 1.12) (712) (.98 − 1.26)
Any partner violence N = 6,564 Column % (n) OR (CI)
28.15 1.03 12.22 1.05 .64 1.09
3.1 .98 6.1 .86 90.8 16 1.01 29 1.16
(6.43) (1.02 − 1.04)∗∗∗ (6.27)∗ (1.05 − 1.06)∗∗∗ (1.03)∗ (1.05 − 1.14)∗∗∗
Past N = 4,394 Mean (SD) OR (CI)
(53) (.69 − 1.38) (103) (.67 − 1.11) (1, 537) (709) (.89 − 1.14) (425) (1.00 − 1.33)∗
Past N = 4,394 Column % (n) OR (CI)
26.98 1.00 10.90 1.02 1.06 1.19
5 1.63 7.9 1.17 87.1 21 1.01 320 1.03
(6.74) (.99 − 1.0) (6.55)∗ (1.01 − 1.03)∗∗∗ (.02)∗ (1.13 − 1.24)∗∗∗
Current N = 2,186 Mean (SD) OR (CI)
(55) (1.15 − 2.30)∗∗ (87) (.90 − 1.53) (957) (467) (.87 − 1.16) (287) (.88 − 1.21)
Current N = 2,186 Column % (n) OR (CI)
Note. ORs and CI s are from unadjusted logistic regressions with no partner violence as the reference group. All variables are categorical except age, number of years sexually active, and parity, which are continuous. Past and current violence are subcategories that make up any IPV, but table shows column percentages, so past and current will not always add up to any partner violence as a row. ∗ p < .05; ∗∗ p < .01; ∗∗∗ p < .001.
.61 (1.01)
26.95 (5.96)
Characteristic
Parity
No IPV N = 6,564 Mean (SD) (ref)
51 (755)
3.2 (86)
Characteristic
Surgical abortion (medical = ref) (1st trimester only n = 4958)
No partner violence N = 6,564 Column % (n) (ref)
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
186
L. Colarossi and G. Dean
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
History of Abortion, Miscarriage, and Partner Violence Multivariate (adjusted) logistic regressions (Table 2) of lifetime history of abortions and miscarriages (with 0 abortions and 0 miscarriages as the reference groups) showed that greater parity was associated with an increased likelihood of both 1 and ≥2 abortions and miscarriages. Additionally, Blacks, Hispanics, and other races all had greater odds of having both 1 and ≥2 abortions and miscarriages compared to Whites, which was also true for problems with birth control. Those with histories of abortion and miscarriage were less likely to self-pay for their current appointment compared to paying with private insurance; we found no difference between those paying with Medicaid and private insurance. Current payment method was not associated with history of abortion and miscarriage; so, this may not be a reliable association, but we wanted to adjust for payment method as a proxy for income in the model. After adjusting for these variables, partner violence was not associated with abortion history. However, women with any history of partner violence had greater odds of 1 (OR = 1.49; CI = 1.30–1.69) and ≥2 miscarriages (OR = 1.73; CI = 1.37–2.20).
Current Abortion and Partner Violence Of the subsample of women receiving an abortion service, 90% had a gestational age of ≤12 weeks; 7% of 13–18 weeks; and 3% of 19–24 weeks. A multivariate analysis (Table 3) showed that younger women and those paying with Medicaid and self-pay compared to private insurance were slightly more likely to be having an abortion at both 13–18 and 19–24 weeks gestational age. We found no associations between gestational age categories and race, parity, or number of years of sexual activity. After adjusting for demographics, women reporting current partner violence were more likely to be having an abortion between 19–24 weeks (OR = 1.62; CI = 1.14–2.30), but not between 13–18 weeks gestation. We found no association between partner violence and return for post-abortion follow-up visit. For the subsample of first trimester abortion patients, women of younger age and greater parity were slightly more likely to choose surgical over medical termination. Those paying with Medicaid and of “other race” were more likely to choose surgical termination compared to those with private insurance and Whites. After adjusting for these demographics, women reporting past partner violence were slightly more likely to choose a surgical procedure than a medical one (OR = 1.16, CI = 1.00–1.33). Those with current violence were no more likely to choose one of these procedures than the other, compared to those with no violence.
187
2.82 (2.62–3.03)∗∗ 3.01 (2.58–3.50)∗∗ 2.71 (2.31–3.17)∗∗ 1.88 (1.57–2.25)∗∗ − .98 (.79–1.22) .38 (.30–.47)∗∗ − 2.10 (1.85–2.38)∗∗
1.64 (1.56 − 1.81)∗∗ 1.59 (1.40 − 1.81)∗∗ 1.72 (1.50 − 1.96)∗∗ 1.44 (1.24 − 1.66)∗∗ – .91 (.74 − 1.11) .44 (.36 − .52)∗∗ – 1.67 (1.48 − 1.89)∗∗ 1.09 (.99–1.21) p = .07
.87 (.85–.89)∗ 1.18 (1.15–1.20)∗∗
.92 (.89 − .93)∗∗ 1.06 (1.04 − 1.08)∗∗
.99 (.90 − 1.08)
≥2 n = 4,052
1 n = 3,635
1.25 (.75–2.07) .77 (.44–1.35) − 1.85 (1.46–2.35)∗∗ 1.73 (1.37–2.20)∗∗
1.49 (1.30–1.69)∗∗
1.91 (1.2–2.96)∗∗ 2.07 (1.33–3.21)∗∗ 1.56 (.92–2.65) −
2.13 (1.95–2.32)∗∗
.86 (.812–.91)∗∗ 1.24 (1.17–1.31)∗∗
≥2 n = 361
.86 (.65–1.13) .73 (.54–.98)∗ − 1.54 (1.33–1.78)∗∗
2.09 (1.65–2.64)∗∗ 1.89 (1.48–2.40)∗∗ 1.62 (1.22–2.14)∗∗ −
2.17 (2.04–2.31)∗∗
.95 (.92–.98)∗∗ 1.07 (1.03–1.10)∗∗
1 n = 1,225
Miscarriage number N = 13,084
Note. ORs and CI s are from logistic regressions with 0 abortions and 0 miscarriages as the reference groups. All variables are categorical except age, number of years sexually active, and parity, which are continuous. ∗ p < .05; ∗∗ p < .001.
Age (years) Years of sexual activity (SD) Parity (SD) Race/ethnicity Non-Hispanic Black Hispanic Other Non-Hispanic White (ref) Payment method Medicaid `Self-pay Commercial (ref) Birth control problems (ref = no) Any partner violence (ref = none)
Characteristic
Abortion number N =13,084
TABLE 2 Adjusted Odds Ratios and 95% Confidence Intervals for Predictors of Abortion and Miscarriage History
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
188
.89 (.71 − 1.10) 1.09 (.86 − 1.40) – 1.04 (.90 − 1.20) 1.04 (.92 − 1.18) –
1.14 (.67–1.96) 1.07 (.62–1.85) .66 (.32–1.35) – 1.21 (.63–2.34) 1.21 (.58–2.53) – 1.62 (1.14–2.30)∗∗ .99 (.70–1.42) –
1.20 (.78 − 1.81) 1.00 (.66 − 1.54) 1.17 (.72 − 1.91) – 2.04 (1.13 − 3.69)∗ 1.64 (.85 − 3.14) – 1.13 (.86 − 1.47) .86 (.67 − 1.11) –
1.11 (.92 − 1.35) 1.8 (.89 − 1.31) .81 (.65 − 1.02) –
1.07 (.91–1.26) 1.16 (1.00–1.33)∗ –
1.42 (1.13–1.79)∗∗ 1.10 (.85–1.42) –
1.20 (.98–1.48) 1.14 (.93–.1.40) 1.43 (1.11–1.83)∗∗ –
1.12 (1.04–1.19)∗∗∗
.76 (.71 − .81)∗∗∗
1.10 (.92–1.30)
1.12 (.99 − 1.27)
97 (.95–.98)∗∗∗ N/A
SAB n =3,491
MAB vs. SAB N = 4,958
1.07 (1.4 − 1.10)∗∗∗ .97 (.94 − 1.00)∗
.92 (.85–.99)∗ 1.03 (.95–1.11)
.89 (.84 − .94)∗∗∗ 1.04 (.98 − 1.12)
Returning n = 2,298
19–24 weeks n = 194
13–18 weeks n = 377
Follow-up N = 5,607
MAB = medical abortion. SAB = surgical abortion. ORs and CIs are from logistic regressions with reference groups of 312 weeks gestational age, no return for follow-up, and MAB. All variables are categorical except age, number of years sexually active, and parity which are continuous. ∗ p < .05; ∗∗ p < .01; ∗∗∗ p < .001.
Age (years) Years of sexual activity (SD) Parity (SD) Race/ethnicity Non-Hispanic Black Hispanic Other Non-Hispanic White (ref) Payment method Medicaid Self-pay Commercial (ref) Partner violence Current Past None (ref)
Predictor variables
Gestational age N = 5,499
TABLE 3 Adjusted Odds Ratios and 95% Confidence Intervals for Predictors of Abortion Characteristics
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Partner Violence and Abortion Characteristics
189
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
DISCUSSION Data from this study adds to the emerging literature showing partner violence to be associated with pregnancy termination—both by miscarriage and abortion. As an associational study using secondary data, the results present an exploratory picture of partner violence and abortion characteristics of timing, number, method, and follow-up. In particular, these data suggest that partner violence is not associated with ever having an abortion or with first trimester termination timing, but that it creates risk for repeat and later abortions from 19–24 weeks and for repeated miscarriages. This is consistent with earlier research reviewed above that first and earlier gestational age abortions are due to a variety of normative factors in women’s lives, while partner violence can create additional complications for delayed and repeat procedures (Ely & Otis, 2011; Foster & Kimport, 2013). This study highlights the need for more research into the mechanisms of influence by a partner on a woman’s decision to terminate a pregnancy with diverse samples from different settings, such as reproductive health centers, domestic violence organizations, and the general population. Also, more in-depth knowledge is needed about the strategies women use to manage their health in such contexts. Some researchers have begun to delineate partner influences on later gestational age abortion through conflicts about the pregnancy, fear, and control over the decision-making process (Coggins & Bullock, 2003; Foster & Kimport, 2013). We also need further data about whether partners also influence other known reasons for later termination, such as later pregnancy testing, controlling finances or medical care which delay receipt of services, and pregnancy intention, to determine whether partner violence may compound or confound these factors. Nationally, 18% of all pregnancies are electively terminated; and among women having abortions, 44% have had at least one prior abortion (Pazol et al., 2011; Ventura et al., 2012). The vast majority of abortion procedures take place in the first trimester, but 8.5% are performed in the second trimester, which is a lengthier, more expensive, higher risk procedure with less availability of doctors performing the procedure and in fewer geographic areas (Jones & Kooistra, 2011). Both repeat and second trimester abortions carry social stigma and expense for the women needing them (Cockrill, et al., 2013; Norris et al., 2011), but little is known about what would be helpful for prevention of repeat unwanted pregnancies and/or delays in obtaining an abortion. Further, women are often stigmatized after disclosing partner violence, creating an additional set of psychological barriers for receipt of services (Maier, 2012; Thapar-Bjorkert & Morgan, 2010), which intensifies the need for both universal partner screening in health centers as well as provider training for conducting effective screening and responses to disclosures (Colarossi, Breitbart, & Betancourt, 2010b). This study reinforces the need for: (1) universal partner violence screening, (2) provider training on
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
190
L. Colarossi and G. Dean
partner violence as a form of power and control and the ways in which that can impact reproductive health, and (3) targeted follow-up counseling protocols. Follow-up counseling should focus on reproductive health and provide community referrals for more comprehensive domestic violence services. Reproductive health counseling protocols for women experiencing partner violence should include assessment of partner influences on sexual and health care practices, identification of risk-reduction strategies including birth control methods useful for each woman’s unique relationship context, and promotion of preventive health care, such as more frequent testing for STIs and early pregnancy testing with options counseling and information about availability of services. While screening protocols and have been evaluated and widely recommended, evaluation of follow-up interventions in health care settings has only just begun (Miller et al., 2011).
Limitations While the strengths of this study include a large, diverse sample of women obtaining family planning and abortion services, the use of secondary data from medical charts posed several limitations. These data were not originally collected for the purposes of this study but rather for clinical practice needs, and we do not have retrospective measures that could provide greater insight into the specific partner behaviors that might influence the participants’ reproductive health and their related decision making. For example, partner violence may take the form of birth control sabotage or other forms of sexual coercion that would make an unwanted pregnancy more likely. If we interpret abortion as a proxy for unwanted pregnancy, these data suggested associations among partner violence, birth control problems, and unwanted pregnancy. However, we cannot really know what proportion were wanted pregnancies that became unmanageable for the woman due to relationship circumstances or were partner-coerced terminations. Further, the association between partner violence and miscarriage may have been due to assault during a pregnancy or a self-induced termination when a woman perceived that her partner might coerce an unwanted pregnancy to be carried to term by preventing a legal abortion. We also do not know the specifics about birth control problems and whether these were directly related to a partner’s behavior. Other than measurement timing and specificity, reporting bias could have created inaccurate estimates in these analyses. In particular, partner violence and abortion number may have been under-reported due to stigma or fear of disclosure. If this was the case, the effect sizes may be larger in the general population than what we found in this study. The New York City sample may also not be representative of other areas in the United States, as abortion rates are higher and second-trimester abortion services are more accessible. Finally, we may have had inadequate control of confounding because data were missing or not collected.
Partner Violence and Abortion Characteristics
191
CONCLUSION
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
This study adds to the growing literature about the relationship between reproductive control by a partner and reproductive health, particularly in the area of abortion characteristics. Findings support the need for universal screening for partner violence in all reproductive health settings, and follow-up counseling protocols focusing on safety planning for contraceptive methods that will allow a woman as much control and choice as possible over her reproductive health within her unique relationship circumstances.
FUNDING This study was funded by the Society of Family Planning. The opinions expressed in this article do not necessarily reflect those of the Society of Family Planning or Planned Parenthood Federation of America, Inc.
REFERENCES Chavkin, D., and R. E. Gee. 2008. Intimate partner violence in women receiving gynecologic and abortion care. Contraception 78(2):179. Cockrill, K. U. D., U. D. Upadhyay, J. Turan, and D. Greene Foster. 2013. The stigma of having an abortion: Development of a scale and characteristics of women experiencing abortion stigma. Perspect Sex Reprod H 45(2):79–88. Coggins, M. and L. F. C. Bullock. 2003. The wavering line in the sand: The effects of domestic violence and sexual coercion. Issues Ment Health Nurs 24:723–38. Coker, A. 2007. Does physical and intimate partner violence affect sexual health?: A systematic review. Trauma Violence Abus 8(2):149–77. Colarossi, L. G., V. Breitbart, and G. Betancourt. 2010a. Screening for intimate partner violence in reproductive health settings: An evaluation study. Women Health 50(4):1–14. Colarossi, L. G., V. Breitbart, and G. Betancourt. 2010b. Barriers to screening for intimate partner violence: A mixed-methods study of providers in family planning clinics. Perspect Sex Reprod H 42(4):236–43. Ely, G. E., and M. D. Otis. 2011. An examination of intimate partner violence and psychological stressors in adult abortion patients. J Interpers Violence 26:3248–66. Finer, L. B., L. A. Frohwirth, L. A. Dauphinee, S. Singh, and A. Moore. 2006. Timing of steps and reasons for delays in obtaining abortions in the United States. Contraception 74(4):334–44. Fisher, W. A., and S. Singh. 2005. Characteristics of women undergoing repeat inducted abortion. Can Med Assoc J 172(5):637–41 Gazmararian, J. A., R. Petersen, A. M. Spitz, M. M. Goodwin, L. E. Saltzman, and J. S. Marks. 2000. Violence and reproductive health: current knowledge and future research directions. Matern Child Heal J 4(2):79–84.
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
192
L. Colarossi and G. Dean
Greene Foster, D., and K. Kimport. 2013. Who seeks abortions at or after 20 weeks? Perspect Sex Reprod H 45(4):210–8. Heise, L. L. 1998. Violence against women: An integrated ecological framework, Violence Against Wom 4(3):262–90. Janeski, J. L. 2004. Pregnancy and domestic violence: A review of the literature. Trauma, Violence, Abuse 5(1):47–64. Jones, R. K., and L. B. Finer. 2012. Who has second-trimester abortions in the United States? Contraception 85:544–51. Jones, R. K., and K. Kooistra. 2011. Abortions incidence and access to services in the United States, 2008. Perspect Sex Reprod H 43(1):41–50. Jones, R. K., A. M. Moore, and L. F. Frowirth. 2010. Perceptions of male knowledge and support among U.S. women obtaining abortions. Women Health Iss 21(2):117–23. Maier, S. L. 2012. The complexity of victim-questioning attitudes by rape victim advocates: Exploring some gray areas. Violence Against Wom 18(2):1413–34. Miller, E., M. R. Decker, H. L. McCauley, D. J. Tancredi, R. R. Levenson, J. Waldman, et al. 2010. Pregnancy coercion, intimate partner violence and unintended pregnancy. Contraception 81:316–22. Miller, E., M. R. Decker, H. L. McCauley, D. J. Tancredi, R. R. Levenson, J. Waldman, et al. 2011. A family planning clinic partner violence intervention to reduce risk associated with reproductive coercion. Contraception 83:274–80. Moore, A. M., L. Frohwirth, and E. Miller. 2010. Male reproductive control of women who have experienced intimate partner violence in the United States. Soc Sci Med 70(11):1737–44. Norris, A., D. Bessett, J. R. Steinberg, M. L. Kavanaugh, S. De Zordo, and D. Becker. 2011. Abortion stigma: A reconceptualization of constituents, causes, and consequences. Women Health Iss 21(3):S49–54. ODonnell, L., G. Agronick, R. Duran, U. A. Nyint, and A. Stueve. 2009. Intimate partner violence among economically disadvantaged young adult women: associations with adolescent risk-taking and pregnancy experiences. Perspect Sex Reprod H 41(2):84–91. Pazol, K., S. B. Zane, L. R. Hall, C. Berg, and D. A. Cook. 2011. Abortion surveillanceUnited States, 2008. MMWR 60(5):1–41. Plichta, S. B., and C. Abraham. 1996. Violence and gynecological health in women
Women & Health Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/wwah20
Partner Violence and Abortion Characteristics a
Lisa Colarossi PhD LCSW & Gillian Dean MD MPH
a
a
Planned Parenthood of New York City, New York, New York, USA Accepted author version posted online: 28 Feb 2014.Published online: 08 Apr 2014.
To cite this article: Lisa Colarossi PhD LCSW & Gillian Dean MD MPH (2014) Partner Violence and Abortion Characteristics, Women & Health, 54:3, 177-193, DOI: 10.1080/03630242.2014.883662 To link to this article: http://dx.doi.org/10.1080/03630242.2014.883662
PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://www.tandfonline.com/page/termsand-conditions
Women & Health, 54:177–193, 2014 Copyright © Taylor & Francis Group, LLC ISSN: 0363-0242 print/1541-0331 online DOI: 10.1080/03630242.2014.883662
Partner Violence and Abortion Characteristics LISA COLAROSSI, PhD, LCSW and GILLIAN DEAN, MD, MPH
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Planned Parenthood of New York City, New York, New York, USA
We conducted a retrospective cohort study using randomly selected medical charts of women reporting a history of partner violence and women with no history of partner violence at the time of a family planning or abortion appointment ( n = 6,564 per group). We analyzed lifetime history of partner violence for odds of lifetime history of abortion and miscarriage number, and birth control problems. To more closely match timing, we analyzed a subsample of 2,186 women reporting current violence versus not at the time of an abortion appointment for differences in gestational age, medical versus surgical method choice, and return for follow-up visit. After adjusting for years at risk and demographic characteristics, women with a past history of partner violence were not more likely to have ever had one abortion, but they were more likely to have had problems with birth control, repeat abortions, and miscarriages than women with no history of violence. Women with current partner violence were also more likely to be receiving an abortion at a later gestational age. We found no differences between the groups in return for abortion follow-up visit or choice of surgical versus medication abortion. Findings support screening for the influence of partner violence on reproductive health and related safety planning. KEYWORDS control
abortion, partner violence, miscarriage, birth
INTRODUCTION More than two decades of research shows associations between partner violence and detrimental sexual and reproductive health outcomes including Received June 5, 2013; revised January 3, 2014; accepted January 9, 2014. Address correspondence to Lisa Colarossi, PhD, LCSW, Planned Parenthood of New York City, 26 Bleecker Street, New York, NY 10012. E-mail: [email protected] 177
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
178
L. Colarossi and G. Dean
lower use of condoms and other contraceptives; higher rates of sexually transmitted infections (STIs), HIV, urinary tract infections, unintended and teen pregnancies; poor birth outcomes; and miscarriage due to physical assault (Coker, 2007; Gazmararian et al., 2000; Plichta & Abraham, 1996; Silverman et al., 2006). In particular, pregnancy is associated with first incidences of partner violence, as well as increased severity (Janeski, 2004; Sagrestano et al., 2004); and pregnancy termination is more likely for women experiencing partner conflicts and violence (Coker, 2007; Ely & Otis, 2011; Foster & Kimport, 2013). However, little is known about whether partner violence is related to abortion characteristics such as method, timing, number, and follow-up. Abortion characteristics have mostly been studied for their association with demographic factors, such as age, race, income, education, marital status, and number of children (Jones & Finer, 2012; Pazol et al., 2011). Similar demographic characteristics are associated with partner violence as well (Rennison & Welchans, 2003; Tjaden & Theonnes, 2000), which may have unique influences on abortion characteristics. New theoretical models propose a number of psychological, social, and physical pathways by which partner violence affects abortion and other reproductive health outcomes (Heise, 1998; Moore, Frohwirth, & Miller, 2010). One important pathway is through reproductive control by a partner, defined as interference with a woman’s ability to make independent/autonomous decisions about her reproduction. Reproductive control can be inflicted economically, emotionally, and/or physically. Characteristics include partner enforcement of his sexual and reproductive intentions in direct conflict with or without consideration of the woman’s intentions via coercion, intimidation, threats, and/or physical assault (Moore et al., 2010). Recent research has provided evidence for the impact of partner refusal to withdraw during sex and/or to use a condom, sabotaging birth control methods, and sexual assault on reproductive outcomes from STIs and HIV transmission to unintended pregnancy and abortion (Ely & Otis, 2011; Moore et al., 2010; Miller et al., 2010; ODonnell et al., 2009; Saftlas et al., 2010; Silverman et al., 2011; Thiel de Bocanegra et al., 2010; Williams, Larsen, & McCloskey, 2008). The few studies that have analyzed partner violence as a variable in abortion research have estimated the proportion of partner violence among women seeking abortion from 11% to 40% (Chavkin & Gee, 2008; Saftlas et al., 2010; Woo, Fine, & Goetzl, 2005). Women experiencing partner violence and fear of negotiating pregnancy decisions with their partner are more likely to have had an abortion and at a later gestational age (Finer et al., 2006; Foster & Kimport, 2013; Jones & Finer, 2012; Jones, Moore, & Frowirth, 2010; Williams & Brackley, 2009), as well as repeat abortions (Ely & Otis, 2011; Fisher & Singh, 2005; ODonnell et al., 2009). Men perpetrating partner violence are more likely to be involved in partner conflicts about whether or not to terminate a pregnancy and to report repeat abortions among their
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Partner Violence and Abortion Characteristics
179
partners (Silverman et al., 2010). Many of these studies have had small sample sizes and have not examined various abortion characteristics within the same study. Our study aim was to compare a large sample of women reporting past and current partner violence to women who did not for differences in a variety of abortion characteristics, after adjusting for some demographic characteristics, years of sexual activity, parity, and problems with birth control. We conducted a retrospective chart review of women receiving abortion and family planning services at three large health centers in New York City. First, we compared women reporting any lifetime history of partner violence to those with no history for differences in lifetime pregnancy outcomes of live births and elective and spontaneous terminations. Second, we analyzed a sub-sample of women obtaining current abortions to compare women reporting current partner violence to those with no current violence for differences in gestational age, medical versus surgical method choice, and return for follow-up visit. In this way we were able to analyze models with closer time-ordering, with lifetime versus current experiences of partner violence and abortion.
METHOD Design and Sampling We conducted a retrospective cohort study using data from medical charts of women receiving a family planning or abortion service over a one-year period (June 1, 2010 to May 31, 2011) at three reproductive health care centers in the Bronx, Brooklyn, and Manhattan in New York City. Family planning services included comprehensive gynecological exams, visits for contraception, pregnancy testing, and STI and HIV testing and treatment. Abortion visits included medical and surgical abortions between 4 and 24 weeks gestational age. From this one-year sampling frame, we randomly selected equal numbers of charts for women (1) reporting any history of partner violence (current or past; N = 6,564) and (2) reporting no partner violence (N = 6,564). An individual woman may have had more than one visit during 2010–2011, and therefore we selected the sample by a singular patient identification and appointment date through computer randomization of electronic medical record data. Information technology staff downloaded de-identified data from electronic medical records into a Micosoft Excel format, which researchers then converted to an IBM SPSS version 20 database for statistical analysis. We received an Institutional Review Board exemption from Mt. Sinai School of Medicine, New York, NY for the use of de-identified medical data in research. Lifetime history of partner violence, abortion number, and miscarriage number were self-reported on the medical history forms of women receiving
180
L. Colarossi and G. Dean
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
reproductive health services. To narrow the temporality of associations among current partner violence and current abortion outcomes, we conducted a sub-analysis of women currently receiving an abortion service from 4 to 24 weeks gestational age from the sample above (N = 5,609; i.e., removing the family planning visits). For these visits, health center staff recorded gestational age and return for post-abortion follow-up visit, which we analyzed for relative risk between women who self-reported current partner violence, past partner violence (but no current), and no partner violence ever. Finally, for women receiving a first trimester abortion (N = 4,958), we analyzed differences in choice of medication versus surgical termination by partner violence group.
Measures DEMOGRAPHIC
CHARACTERISTICS
Demographic characteristics of age, age of sexual debut, race/ethnicity, and payment method were derived from patients’ self-reports. Race/ethnicity was coded into four categories: Black (non-Hispanic), Hispanic, Other, and White (non-Hispanic). We coded payment method into three categories: Medicaid, self-pay, and commercial/private insurance. We analyzed age as a continuous variable in years. We created the variable for number of years at sexual risk by subtracting age of sexual debut from current age and analyzed as a continuous variable in years. We did not analyze other typical demographic characteristics because the electronic medical record had a large amount of missing data for employment status, marital status, and education as many patients do not to report this information and are not required to do so to receive services. PARTNER
VIOLENCE
Partner violence was derived from patients’ self-reported answers to empirically based questions about partner violence in the past year and ever (Colarossi, Breitbart, and Betancourt, 2010a; Zeitler et al., 2006). Three questions asked about partner violence in the past year that included being: (1) hit, kicked, or slapped; (2) threatened or frightened; and (3) forced to have sex. Two questions asked if prior to the last year, any partner had ever (1) hit, kicked, or slapped them and (2) sexually assaulted them. Patients answered these questions in writing on the medical history form, and were asked again verbally by health center staff. We created a dichotomous yes/no variable for any partner violence history by coding yes if an affirmative answer was given to any of the six questions (past and/or current violence) or not. We also created a variable for current partner violence by coding the data into three categories: current partner violence experienced in the past
Partner Violence and Abortion Characteristics
181
year, past partner violence but not in the last year, and no partner violence history. BIRTH
CONTROL PROBLEMS
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Patients reported birth control problems on the medical history form in response to the question, Have you ever had problems with your birth control method? We dichotomously coded the answers into yes/no. No detailed information about the type of problem was recorded on the medical history form, but rather was discussed in counseling with a health care provider. Thus, additional data were not available from electronic medical records. HISTORY
OF PREGNANCY OUTCOMES
History of pregnancy outcomes was derived from self-reported information on the medical history form. We coded number of lifetime miscarriages (spontaneous terminations) and abortions (elective terminations) into three categories: 0, 1, and ≥2 for each outcome. We used number of live births was as a continuous covariate in the prediction of miscarriage and abortion outcomes, which was a measure of parity and not of how many children a woman was currently parenting. ABORTION
CHARACTERISTICS
Health center staff recorded abortion characteristics for women receiving an abortion service on the selected appointment date in in the medical chart. We coded gestational age (via pre-procedure ultrasound) into three categories: ≤12 weeks, 13–18 weeks, and 19–24 weeks, and patient return for a post-abortion follow-up visit as yes/no. We dichotomously coded abortion method as medical or surgical for those women up to 9 weeks gestation who were allowed to choose their termination method after being counseling on their options.
Data Analysis First, we examined bivariate associations between partner violence categories (none any, past only, and current) and demographic characteristics (age, race, and payment method), years of sexual activity, birth control problems, pregnancy outcomes (number of live births, miscarriages, and abortions), and abortion characteristics (gestational timing, method, and return for follow-up). Second, we analyzed separate ordinal logistic regression models for the odds of lifetime number of abortions and miscarriages (both categorized as none, one, or ≥2) by any partner violence history
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
182
L. Colarossi and G. Dean
(yes/no), adjusting for age, number of years of sexual activity, parity, race, payment method, and birth control problems. Each demographic characteristic (age, race, years sex, parity, payment, and birth control problems) was significantly associated with partner violence; therefore, all were included in the regression model to remove the variance associated with the dependent variables before estimating the effects of partner violence (i.e., to adjust for the covariance with partner violence). All covariates remained significant related in the multivariate models with no significant interaction effects of IPV with other covariates; so, we tested only main effects of covariates in the final models. Third, we analyzed separate ordinal and binary logistic regression models for the subsample of abortion patients only (i.e., removing cases that were currently receiving a family planning service) for the odds of gestational timing (1–12, 13–18, and 19–24 weeks) and return for follow-up (yes/no) by current partner violence after adjusting for age, number of years sexually active, parity, race, and payment method. Finally, we analyzed a binary logistic regression model for those cases receiving first trimester abortion services for the odds of currently choosing a medical versus a surgical abortion procedure by current partner violence after adjusting for age, race, and payment method (only first trimester pregnancies have the option of a procedure type). We removed history of birth control problems from all final models related to current abortion characteristics because it was not a significant predictor of these variables, nor did we theoretically expect that it would be a predictor of gestational age, termination method, or return for follow-up. We found no significant interaction effects of IPV with the covariates on the dependent variables, so we tested main effects only in the final models reported here. We found that all models had good fit were by virtue of statistically significant -2 log-likelihood statistics showing that the model variables outperformed the null hypothesis, as well as small, non-significance Pearson and Deviance goodness-of-fit statistics. We used SPSS version 20 for data analysis.
RESULTS Demographics and Partner Violence The sample was highly diverse with ages ranging from 14–60 years (Median = 26.4 years), years of sexual activity ranging from 0–45 years (Median = 10 years), and parity ranging from 0–9 (Median = 0). 35% of women identified as black, 29% as Hispanic (White and non-White), 23% white, and 13% other races. Fifty-nine percent of women paid for their current visit with Medicaid, 36% self-paid, and 5% with commercial/private insurance. Of women reporting any partner violence history, 33% reported violence in the past year.
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Partner Violence and Abortion Characteristics
183
Women reporting any partner violence were slightly older (Mean = 27.8 years, Standard Deviation = 6.6 years) than those reporting none (Mean = 26.95 years, Standard Deviation = 5.96 years), as were those reporting past violence only (Mean = 28.1 years, Standard Deviation = 6.43 years; Table 1). Women reporting current violence were the same age as those reporting none (Mean = 27.0 years, Standard Deviation = 6.7 years). This pattern was similar for years of sexual activity and parity, also with statistically significant but very small differences. Black women had a slightly higher odds of reporting any partner violence versus none compared to White women (odd ratio [OR] = 1.13, confidence interval [CI ] = 1.03–1.23), while Hispanic and others were similar to White women. No racial differences was observed in past violence versus no violence, but Black (OR = 1.53, CI = 1.34–1.76), Hispanic (OR = 1.40, CI = 1.21–1.61) and other women (OR = 1.26, CI = 1.06–1.49) had higher odds of reporting current violence compared to White women. Women reporting any (OR = 1.16, CI = 1.0–1.36) and current partner violence (OR = 1.30, CI = 1.34–1.63) were slightly more likely to be paying with Medicaid than commercial insurance. We found no differences between self-pay and commercial insurance by partner violence experience. In the full sample, 18% of women reported ever having problems using birth control; 9% had a lifetime history of one miscarriage, and 3% had ≥2 miscarriages. A total of 41% of women had never had a therapeutic abortion; 28% had one prior abortion, and 31% had ≥2 abortions. Compared to women with no partner violence, women reporting any history of violence, past and current, were all more likely to have experienced problems with birth control and to have had ≥2 miscarriages. Women who reported current partner violence were more likely to have had one prior abortion (OR = 1.70, CI = 1.44–1.99) than those reporting no violence. We found no differences between women who reported past violence and those who reported no violence for having had one prior abortion. However, those reporting past (OR = 1.24; CI = 1.13–1.36) and current violence (OR = 2.38, CI = 1.79–3.16) were both more likely to have had ≥2 abortions in their lifetime compared to no abortions than women who did not report histories of partner violence. For the subsample of women receiving an abortion on the randomly selected appointment date, those reporting current violence (OR = 1.63; CI = 1.15–2.30) were more likely to have an abortion appointment at a gestational age from 19–24 weeks compared to ≤12 weeks than women with no violence. We found no other differences between partner violence categories and gestational age. We also found no differences between partner violence categories and return for a follow-up visit. For those women who were receiving a first trimester abortion, 70% chose a surgical procedure. We found no association between partner violence and choice to have a medical versus a surgical termination procedure.
184 27.8 (1824) 43.8 (2876)
0 (ref)
28.4 (1864)
81.2 (5327)
>=2
0 (ref) Number of abortions 1
11.4 (751)
7.4 (486)
Number of miscarriages 1
>=2
5.5 (362) 40.3 (978)
37.5 (2464)
Commercial (ref) Birth control problems
Self-pay
57 (3738)
22.1 (1529)
13 (965)
Other
Non-Hispanic White (ref) Payment method Medicaid
29 (1864)
35.9 (2206)
Hispanic
Race/ethnicity Non-Hispanic Black
Characteristic
No partner violence N = 6,564 Column % (n) (ref)
27.1 1.08 34.2 1.39 38.7
11.4 1.65 1.8 2.27 86.8
60.5 1.16 34.5 1.01 5 59.7 1.61
33.6 1.13 28.4 1.08 14.7 .93 23.3
(1, 791) (1.0 − 1.19)p= .◦6 (2, 256) (1.28 − 1.15)∗∗∗ (2, 251)
(751) (1.46 − 1.85)∗∗∗ (117) (1.81 − 2.83)∗∗∗ (5, 696)
(3, 988) (1.0 − 1.36)∗ (2, 278) (.86 − 1.18) (332) (1, 450) (1.47 − 1.76)∗∗∗
(2368) (1.03 − 1.23)∗∗ (1916) (.98 − 1.19) (857) (.83 − 1.05) (1457)
Any partner violence N = 6,564 Column % (n) OR (CI)
(2, 550) (.92 − 1.30) (1, 619) (.72 − 1.15) (225) (989) (1.50 − 1.83)∗∗∗
(1509) (.88 − 1.08) (1260) (.87 − 1.07) (552) (.72 − .93)∗∗ (1073)
24.2 .88 33.4 1.24 42.4
(1, 065) (.80 − .97)∗∗ (1, 464) (1.13 − 1.36)∗∗∗ (1, 865)
11.3 (495) 1.63 (1.43 − 1.83) 3.7 (163) 2.23 (1.75 − 2.84)∗∗∗ 85(3, 736)
58.0 1.10 36.8 .91 5.2 40.7 1.66
34.3 .97 28.7 .96 12.6 .81 24.4
Past N = 4,394 Column % (n) OR (CI)
33 1.64 36 1.84 31
11.7 1.70 3.9 2.38 84.4
65.2 1.30 30 .91 4.8 18.8 1.51
38.8 1.53 29.8 1.40 13.9 1.26 17.5
(721) (1.46 − 1.85)∗∗∗ (788) (1.63 − 2.07)∗∗∗ (677)
(255) (1.44 − 1.99) (86) (1.79 − 3.16)∗∗∗ (1, 919)
(1, 425) (1.34 − 1.63)∗ (655) (.72 − 1.15) (106) (457) (1.33 − 1.71)∗∗∗
(849) (1.34 − 1.76)∗∗∗ (652) (1.21 − 1.61)∗∗∗ (303) (1.06 − 1.49)∗∗ (382)
Current N = 2,186 Column % (n) OR (CI)
TABLE 1 Percent Distribution of Demographic Characteristics and Pregnancy Outcomes by Partner Violence with Unadjusted Odds Ratios (OR) and 95% Confidence Intervals (CI)
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
185
6.9 (189) 89.9 (2, 459) 49 (1, 144)
Gestational age 19–24 weeks
13–18 weeks
1–12 weeks (ref) Follow-up (yes)
10.20 (5.79)
Age (years)
Number of years of sexual activity
27.76 1.02 11.78 1.04 .67 1.12
3.9 1.25 6.9 1.00 89.2 51 1.01 49 1.11
(6.56) (1.01 − 1.03)∗∗∗ (6.40)∗ (1.04 − 1.05)∗∗∗ (1.04)∗ (1.09 − 1.16)∗∗∗
Any IPV N = 6,564 Mean (SD) OR (CI)
(109) (.93–1.66) (192) (.81 − 1.23) (2, 499) (1, 180) (.91 − 1.12) (712) (.98 − 1.26)
Any partner violence N = 6,564 Column % (n) OR (CI)
28.15 1.03 12.22 1.05 .64 1.09
3.1 .98 6.1 .86 90.8 16 1.01 29 1.16
(6.43) (1.02 − 1.04)∗∗∗ (6.27)∗ (1.05 − 1.06)∗∗∗ (1.03)∗ (1.05 − 1.14)∗∗∗
Past N = 4,394 Mean (SD) OR (CI)
(53) (.69 − 1.38) (103) (.67 − 1.11) (1, 537) (709) (.89 − 1.14) (425) (1.00 − 1.33)∗
Past N = 4,394 Column % (n) OR (CI)
26.98 1.00 10.90 1.02 1.06 1.19
5 1.63 7.9 1.17 87.1 21 1.01 320 1.03
(6.74) (.99 − 1.0) (6.55)∗ (1.01 − 1.03)∗∗∗ (.02)∗ (1.13 − 1.24)∗∗∗
Current N = 2,186 Mean (SD) OR (CI)
(55) (1.15 − 2.30)∗∗ (87) (.90 − 1.53) (957) (467) (.87 − 1.16) (287) (.88 − 1.21)
Current N = 2,186 Column % (n) OR (CI)
Note. ORs and CI s are from unadjusted logistic regressions with no partner violence as the reference group. All variables are categorical except age, number of years sexually active, and parity, which are continuous. Past and current violence are subcategories that make up any IPV, but table shows column percentages, so past and current will not always add up to any partner violence as a row. ∗ p < .05; ∗∗ p < .01; ∗∗∗ p < .001.
.61 (1.01)
26.95 (5.96)
Characteristic
Parity
No IPV N = 6,564 Mean (SD) (ref)
51 (755)
3.2 (86)
Characteristic
Surgical abortion (medical = ref) (1st trimester only n = 4958)
No partner violence N = 6,564 Column % (n) (ref)
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
186
L. Colarossi and G. Dean
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
History of Abortion, Miscarriage, and Partner Violence Multivariate (adjusted) logistic regressions (Table 2) of lifetime history of abortions and miscarriages (with 0 abortions and 0 miscarriages as the reference groups) showed that greater parity was associated with an increased likelihood of both 1 and ≥2 abortions and miscarriages. Additionally, Blacks, Hispanics, and other races all had greater odds of having both 1 and ≥2 abortions and miscarriages compared to Whites, which was also true for problems with birth control. Those with histories of abortion and miscarriage were less likely to self-pay for their current appointment compared to paying with private insurance; we found no difference between those paying with Medicaid and private insurance. Current payment method was not associated with history of abortion and miscarriage; so, this may not be a reliable association, but we wanted to adjust for payment method as a proxy for income in the model. After adjusting for these variables, partner violence was not associated with abortion history. However, women with any history of partner violence had greater odds of 1 (OR = 1.49; CI = 1.30–1.69) and ≥2 miscarriages (OR = 1.73; CI = 1.37–2.20).
Current Abortion and Partner Violence Of the subsample of women receiving an abortion service, 90% had a gestational age of ≤12 weeks; 7% of 13–18 weeks; and 3% of 19–24 weeks. A multivariate analysis (Table 3) showed that younger women and those paying with Medicaid and self-pay compared to private insurance were slightly more likely to be having an abortion at both 13–18 and 19–24 weeks gestational age. We found no associations between gestational age categories and race, parity, or number of years of sexual activity. After adjusting for demographics, women reporting current partner violence were more likely to be having an abortion between 19–24 weeks (OR = 1.62; CI = 1.14–2.30), but not between 13–18 weeks gestation. We found no association between partner violence and return for post-abortion follow-up visit. For the subsample of first trimester abortion patients, women of younger age and greater parity were slightly more likely to choose surgical over medical termination. Those paying with Medicaid and of “other race” were more likely to choose surgical termination compared to those with private insurance and Whites. After adjusting for these demographics, women reporting past partner violence were slightly more likely to choose a surgical procedure than a medical one (OR = 1.16, CI = 1.00–1.33). Those with current violence were no more likely to choose one of these procedures than the other, compared to those with no violence.
187
2.82 (2.62–3.03)∗∗ 3.01 (2.58–3.50)∗∗ 2.71 (2.31–3.17)∗∗ 1.88 (1.57–2.25)∗∗ − .98 (.79–1.22) .38 (.30–.47)∗∗ − 2.10 (1.85–2.38)∗∗
1.64 (1.56 − 1.81)∗∗ 1.59 (1.40 − 1.81)∗∗ 1.72 (1.50 − 1.96)∗∗ 1.44 (1.24 − 1.66)∗∗ – .91 (.74 − 1.11) .44 (.36 − .52)∗∗ – 1.67 (1.48 − 1.89)∗∗ 1.09 (.99–1.21) p = .07
.87 (.85–.89)∗ 1.18 (1.15–1.20)∗∗
.92 (.89 − .93)∗∗ 1.06 (1.04 − 1.08)∗∗
.99 (.90 − 1.08)
≥2 n = 4,052
1 n = 3,635
1.25 (.75–2.07) .77 (.44–1.35) − 1.85 (1.46–2.35)∗∗ 1.73 (1.37–2.20)∗∗
1.49 (1.30–1.69)∗∗
1.91 (1.2–2.96)∗∗ 2.07 (1.33–3.21)∗∗ 1.56 (.92–2.65) −
2.13 (1.95–2.32)∗∗
.86 (.812–.91)∗∗ 1.24 (1.17–1.31)∗∗
≥2 n = 361
.86 (.65–1.13) .73 (.54–.98)∗ − 1.54 (1.33–1.78)∗∗
2.09 (1.65–2.64)∗∗ 1.89 (1.48–2.40)∗∗ 1.62 (1.22–2.14)∗∗ −
2.17 (2.04–2.31)∗∗
.95 (.92–.98)∗∗ 1.07 (1.03–1.10)∗∗
1 n = 1,225
Miscarriage number N = 13,084
Note. ORs and CI s are from logistic regressions with 0 abortions and 0 miscarriages as the reference groups. All variables are categorical except age, number of years sexually active, and parity, which are continuous. ∗ p < .05; ∗∗ p < .001.
Age (years) Years of sexual activity (SD) Parity (SD) Race/ethnicity Non-Hispanic Black Hispanic Other Non-Hispanic White (ref) Payment method Medicaid `Self-pay Commercial (ref) Birth control problems (ref = no) Any partner violence (ref = none)
Characteristic
Abortion number N =13,084
TABLE 2 Adjusted Odds Ratios and 95% Confidence Intervals for Predictors of Abortion and Miscarriage History
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
188
.89 (.71 − 1.10) 1.09 (.86 − 1.40) – 1.04 (.90 − 1.20) 1.04 (.92 − 1.18) –
1.14 (.67–1.96) 1.07 (.62–1.85) .66 (.32–1.35) – 1.21 (.63–2.34) 1.21 (.58–2.53) – 1.62 (1.14–2.30)∗∗ .99 (.70–1.42) –
1.20 (.78 − 1.81) 1.00 (.66 − 1.54) 1.17 (.72 − 1.91) – 2.04 (1.13 − 3.69)∗ 1.64 (.85 − 3.14) – 1.13 (.86 − 1.47) .86 (.67 − 1.11) –
1.11 (.92 − 1.35) 1.8 (.89 − 1.31) .81 (.65 − 1.02) –
1.07 (.91–1.26) 1.16 (1.00–1.33)∗ –
1.42 (1.13–1.79)∗∗ 1.10 (.85–1.42) –
1.20 (.98–1.48) 1.14 (.93–.1.40) 1.43 (1.11–1.83)∗∗ –
1.12 (1.04–1.19)∗∗∗
.76 (.71 − .81)∗∗∗
1.10 (.92–1.30)
1.12 (.99 − 1.27)
97 (.95–.98)∗∗∗ N/A
SAB n =3,491
MAB vs. SAB N = 4,958
1.07 (1.4 − 1.10)∗∗∗ .97 (.94 − 1.00)∗
.92 (.85–.99)∗ 1.03 (.95–1.11)
.89 (.84 − .94)∗∗∗ 1.04 (.98 − 1.12)
Returning n = 2,298
19–24 weeks n = 194
13–18 weeks n = 377
Follow-up N = 5,607
MAB = medical abortion. SAB = surgical abortion. ORs and CIs are from logistic regressions with reference groups of 312 weeks gestational age, no return for follow-up, and MAB. All variables are categorical except age, number of years sexually active, and parity which are continuous. ∗ p < .05; ∗∗ p < .01; ∗∗∗ p < .001.
Age (years) Years of sexual activity (SD) Parity (SD) Race/ethnicity Non-Hispanic Black Hispanic Other Non-Hispanic White (ref) Payment method Medicaid Self-pay Commercial (ref) Partner violence Current Past None (ref)
Predictor variables
Gestational age N = 5,499
TABLE 3 Adjusted Odds Ratios and 95% Confidence Intervals for Predictors of Abortion Characteristics
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
Partner Violence and Abortion Characteristics
189
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
DISCUSSION Data from this study adds to the emerging literature showing partner violence to be associated with pregnancy termination—both by miscarriage and abortion. As an associational study using secondary data, the results present an exploratory picture of partner violence and abortion characteristics of timing, number, method, and follow-up. In particular, these data suggest that partner violence is not associated with ever having an abortion or with first trimester termination timing, but that it creates risk for repeat and later abortions from 19–24 weeks and for repeated miscarriages. This is consistent with earlier research reviewed above that first and earlier gestational age abortions are due to a variety of normative factors in women’s lives, while partner violence can create additional complications for delayed and repeat procedures (Ely & Otis, 2011; Foster & Kimport, 2013). This study highlights the need for more research into the mechanisms of influence by a partner on a woman’s decision to terminate a pregnancy with diverse samples from different settings, such as reproductive health centers, domestic violence organizations, and the general population. Also, more in-depth knowledge is needed about the strategies women use to manage their health in such contexts. Some researchers have begun to delineate partner influences on later gestational age abortion through conflicts about the pregnancy, fear, and control over the decision-making process (Coggins & Bullock, 2003; Foster & Kimport, 2013). We also need further data about whether partners also influence other known reasons for later termination, such as later pregnancy testing, controlling finances or medical care which delay receipt of services, and pregnancy intention, to determine whether partner violence may compound or confound these factors. Nationally, 18% of all pregnancies are electively terminated; and among women having abortions, 44% have had at least one prior abortion (Pazol et al., 2011; Ventura et al., 2012). The vast majority of abortion procedures take place in the first trimester, but 8.5% are performed in the second trimester, which is a lengthier, more expensive, higher risk procedure with less availability of doctors performing the procedure and in fewer geographic areas (Jones & Kooistra, 2011). Both repeat and second trimester abortions carry social stigma and expense for the women needing them (Cockrill, et al., 2013; Norris et al., 2011), but little is known about what would be helpful for prevention of repeat unwanted pregnancies and/or delays in obtaining an abortion. Further, women are often stigmatized after disclosing partner violence, creating an additional set of psychological barriers for receipt of services (Maier, 2012; Thapar-Bjorkert & Morgan, 2010), which intensifies the need for both universal partner screening in health centers as well as provider training for conducting effective screening and responses to disclosures (Colarossi, Breitbart, & Betancourt, 2010b). This study reinforces the need for: (1) universal partner violence screening, (2) provider training on
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
190
L. Colarossi and G. Dean
partner violence as a form of power and control and the ways in which that can impact reproductive health, and (3) targeted follow-up counseling protocols. Follow-up counseling should focus on reproductive health and provide community referrals for more comprehensive domestic violence services. Reproductive health counseling protocols for women experiencing partner violence should include assessment of partner influences on sexual and health care practices, identification of risk-reduction strategies including birth control methods useful for each woman’s unique relationship context, and promotion of preventive health care, such as more frequent testing for STIs and early pregnancy testing with options counseling and information about availability of services. While screening protocols and have been evaluated and widely recommended, evaluation of follow-up interventions in health care settings has only just begun (Miller et al., 2011).
Limitations While the strengths of this study include a large, diverse sample of women obtaining family planning and abortion services, the use of secondary data from medical charts posed several limitations. These data were not originally collected for the purposes of this study but rather for clinical practice needs, and we do not have retrospective measures that could provide greater insight into the specific partner behaviors that might influence the participants’ reproductive health and their related decision making. For example, partner violence may take the form of birth control sabotage or other forms of sexual coercion that would make an unwanted pregnancy more likely. If we interpret abortion as a proxy for unwanted pregnancy, these data suggested associations among partner violence, birth control problems, and unwanted pregnancy. However, we cannot really know what proportion were wanted pregnancies that became unmanageable for the woman due to relationship circumstances or were partner-coerced terminations. Further, the association between partner violence and miscarriage may have been due to assault during a pregnancy or a self-induced termination when a woman perceived that her partner might coerce an unwanted pregnancy to be carried to term by preventing a legal abortion. We also do not know the specifics about birth control problems and whether these were directly related to a partner’s behavior. Other than measurement timing and specificity, reporting bias could have created inaccurate estimates in these analyses. In particular, partner violence and abortion number may have been under-reported due to stigma or fear of disclosure. If this was the case, the effect sizes may be larger in the general population than what we found in this study. The New York City sample may also not be representative of other areas in the United States, as abortion rates are higher and second-trimester abortion services are more accessible. Finally, we may have had inadequate control of confounding because data were missing or not collected.
Partner Violence and Abortion Characteristics
191
CONCLUSION
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
This study adds to the growing literature about the relationship between reproductive control by a partner and reproductive health, particularly in the area of abortion characteristics. Findings support the need for universal screening for partner violence in all reproductive health settings, and follow-up counseling protocols focusing on safety planning for contraceptive methods that will allow a woman as much control and choice as possible over her reproductive health within her unique relationship circumstances.
FUNDING This study was funded by the Society of Family Planning. The opinions expressed in this article do not necessarily reflect those of the Society of Family Planning or Planned Parenthood Federation of America, Inc.
REFERENCES Chavkin, D., and R. E. Gee. 2008. Intimate partner violence in women receiving gynecologic and abortion care. Contraception 78(2):179. Cockrill, K. U. D., U. D. Upadhyay, J. Turan, and D. Greene Foster. 2013. The stigma of having an abortion: Development of a scale and characteristics of women experiencing abortion stigma. Perspect Sex Reprod H 45(2):79–88. Coggins, M. and L. F. C. Bullock. 2003. The wavering line in the sand: The effects of domestic violence and sexual coercion. Issues Ment Health Nurs 24:723–38. Coker, A. 2007. Does physical and intimate partner violence affect sexual health?: A systematic review. Trauma Violence Abus 8(2):149–77. Colarossi, L. G., V. Breitbart, and G. Betancourt. 2010a. Screening for intimate partner violence in reproductive health settings: An evaluation study. Women Health 50(4):1–14. Colarossi, L. G., V. Breitbart, and G. Betancourt. 2010b. Barriers to screening for intimate partner violence: A mixed-methods study of providers in family planning clinics. Perspect Sex Reprod H 42(4):236–43. Ely, G. E., and M. D. Otis. 2011. An examination of intimate partner violence and psychological stressors in adult abortion patients. J Interpers Violence 26:3248–66. Finer, L. B., L. A. Frohwirth, L. A. Dauphinee, S. Singh, and A. Moore. 2006. Timing of steps and reasons for delays in obtaining abortions in the United States. Contraception 74(4):334–44. Fisher, W. A., and S. Singh. 2005. Characteristics of women undergoing repeat inducted abortion. Can Med Assoc J 172(5):637–41 Gazmararian, J. A., R. Petersen, A. M. Spitz, M. M. Goodwin, L. E. Saltzman, and J. S. Marks. 2000. Violence and reproductive health: current knowledge and future research directions. Matern Child Heal J 4(2):79–84.
Downloaded by [Lancaster University Library] at 03:16 21 October 2014
192
L. Colarossi and G. Dean
Greene Foster, D., and K. Kimport. 2013. Who seeks abortions at or after 20 weeks? Perspect Sex Reprod H 45(4):210–8. Heise, L. L. 1998. Violence against women: An integrated ecological framework, Violence Against Wom 4(3):262–90. Janeski, J. L. 2004. Pregnancy and domestic violence: A review of the literature. Trauma, Violence, Abuse 5(1):47–64. Jones, R. K., and L. B. Finer. 2012. Who has second-trimester abortions in the United States? Contraception 85:544–51. Jones, R. K., and K. Kooistra. 2011. Abortions incidence and access to services in the United States, 2008. Perspect Sex Reprod H 43(1):41–50. Jones, R. K., A. M. Moore, and L. F. Frowirth. 2010. Perceptions of male knowledge and support among U.S. women obtaining abortions. Women Health Iss 21(2):117–23. Maier, S. L. 2012. The complexity of victim-questioning attitudes by rape victim advocates: Exploring some gray areas. Violence Against Wom 18(2):1413–34. Miller, E., M. R. Decker, H. L. McCauley, D. J. Tancredi, R. R. Levenson, J. Waldman, et al. 2010. Pregnancy coercion, intimate partner violence and unintended pregnancy. Contraception 81:316–22. Miller, E., M. R. Decker, H. L. McCauley, D. J. Tancredi, R. R. Levenson, J. Waldman, et al. 2011. A family planning clinic partner violence intervention to reduce risk associated with reproductive coercion. Contraception 83:274–80. Moore, A. M., L. Frohwirth, and E. Miller. 2010. Male reproductive control of women who have experienced intimate partner violence in the United States. Soc Sci Med 70(11):1737–44. Norris, A., D. Bessett, J. R. Steinberg, M. L. Kavanaugh, S. De Zordo, and D. Becker. 2011. Abortion stigma: A reconceptualization of constituents, causes, and consequences. Women Health Iss 21(3):S49–54. ODonnell, L., G. Agronick, R. Duran, U. A. Nyint, and A. Stueve. 2009. Intimate partner violence among economically disadvantaged young adult women: associations with adolescent risk-taking and pregnancy experiences. Perspect Sex Reprod H 41(2):84–91. Pazol, K., S. B. Zane, L. R. Hall, C. Berg, and D. A. Cook. 2011. Abortion surveillanceUnited States, 2008. MMWR 60(5):1–41. Plichta, S. B., and C. Abraham. 1996. Violence and gynecological health in women
