If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Using the social determinants framework as a guide, this study sought to understand correlates of postabortion contraceptive use at the individual, family and abortion service delivery levels.
This prospective study assessed correlates of contraceptive use 4 months postabortion and timing of initiation using a facility-based sample of 398 abortion clients who selected pills, condoms, injectables or no method immediately following the procedure. We measured potential correlates immediately following abortion, inclusive of spontaneous or induced abortion, and assessed contraceptive use outcomes 4 months postabortion. Multivariable logistic regression models identified correlates at each level. Potential individual level correlates included contraceptive and abortion history and fertility intentions; family correlates included intimate partner violence (IPV), discordance in fertility intentions and household decision-making; and service delivery correlates included procedure type and postabortion contraceptive counseling.
Reported contraceptive use 4 months postabortion was high (85.4%). Contraceptive use at the index pregnancy (resulting in abortion) was the primary correlate of contraceptive use 4 months postabortion (adjusted odds ratio=2.9; 95% confidence interval: 1.5–5.9). Delayed contraceptive initiation was more common among women who reported past year IPV (36.8% vs. 19.5%; p=.03) particularly with spousal accompaniment for abortion, those in relationships with discordant fertility intentions (44.4% vs. 21.9%; p=.04) and those receiving medication abortion (56.7%) or dilation and curettage (57.1%), compared to manual vacuum aspiration (12.6%; p<.01).
Contraceptive use at the index pregnancy was the primary correlate of contraceptive use 4 months postabortion. Abortion procedure type and relationship dynamics were correlated with delayed postabortion contraceptive initiation. Women who reported IPV delayed initiation when accompanied by their spouse for abortion.
Postabortion contraceptive counseling should assess previous use patterns and provide information on using contraception effectively. Delayed initiation among women reporting IPV could be addressed through comprehensive, confidential counseling that includes violence screening, support for contraceptive initiation and offer of woman-controlled methods.
Fertility can return within 2 weeks of an abortion procedure, and the World Health Organization (WHO) recommends immediate contraceptive provision, on the day of the procedure, to reduce subsequent unwanted pregnancy [
]. However, contraceptive behavior is complex, and provision of contraception may not result in effective contraceptive use, as individual characteristics and an individual's social environment influence behavior [
]. Postabortion contraceptive provision, inclusive of contraceptives provided immediately following MR or PAC procedures, varies considerably based on facility characteristics and method availability [
] are associated with contraceptive continuation. At the family level, women's power within the family is important, especially in rural Bangladesh where husbands' disapproval and dissatisfaction are cited as reasons for contraceptive discontinuation [
]. In this setting, it is more relevant to consider immediate as compared to delayed initiation to understand the determinants affecting implementation of WHO guidelines.
The present study assessed correlates of contraceptive use 4 months postabortion and timing of initiation (immediate or delayed) at the individual, family and abortion service delivery levels among public sector abortion clients in Bangladesh.
2. Material and methods
This prospective study used data from a parent study, which employed a stratified one-stage cluster sampling approach to select 498 women from 16 government and nongovernment organization (NGO) facilities receiving an intervention that trained providers in woman-centered abortion service provision. The parent study stratified facilities by type (primary, secondary, tertiary and NGO clinics) and randomly selected 16 using probability proportional to size sampling within facility-type strata. All facilities in the sample offered both MR services for women presenting with ongoing pregnancy and PAC services for women presenting with incomplete abortion. By policy, providers did not confirm pregnancy prior to an MR procedure [
], but this study considered MR to be equivalent to induced abortion. Providers most commonly used manual vacuum aspiration (MVA) for both MR and PAC cases. Dilation and curettage (D&C) is not a WHO-approved abortion procedure, but some senior obstetrician–gynecologists still routinely used the procedure. In addition, some second trimester PAC cases required D&C because facilities primarily procured small MVA cannulae for MR procedures, which could not be used for PAC clients presenting at later gestational ages. At the time of data collection, medication abortion (MA) was newly available, primarily in the larger health facilities. All facilities provided at least two contraceptive methods, typically oral contraceptive pills and injectables. Inclusion criteria for participants were 18–49 years of age and selected pills, injectables or condoms as a postabortion contraceptive method or selected no method. The parent study focused on women who selected pills, injectables or condoms because they were of particular interest to the Bangladeshi government as they made up the majority of postabortion contraceptive users and because methods such as intrauterine devices (IUDs), implants and sterilization were not available to abortion clients in most government health facilities due to inadequate training and supplies [
]. Thus, women who selected IUDs, implants or sterilization immediately following abortion (approximately 14% of abortion clients, most seen in specialized NGO-run abortion clinics) were ineligible for enrollment in the parent study. Women completed an interviewer-administered survey at the health facility after recovering from their abortion procedure and a follow-up survey 4 months later at a location of their choosing.
The present analysis included all abortion clients who did not wish to become pregnant again soon, including miscarriage clients, as we expected postabortion contraceptive behavior to reflect current pregnancy intentions rather than abortion treatment type. Therefore, we included in the analysis miscarriage clients who intended to delay or limit childbearing. We excluded women from the analysis if on the day of the procedure they intended pregnancy in the next 4 months or, if at the 4-month follow-up, they intended pregnancy in the next month. We also excluded women if they were pregnant at the 4-month follow-up, missing data on contraceptive use at the 4-month follow-up or lost to follow-up.
The primary outcome of interest was contraceptive use 4 months postabortion, which included use of pills, condoms, injectables, implants, IUDs or sterilization. Even though women who selected IUDs, implants or sterilization immediately following abortion were ineligible for enrollment in the parent study, we included in the analysis women who initiated use of these methods over the 4-month follow-up period. We evaluated timing of postabortion contraceptive initiation (immediate or delayed), the secondary outcome of interest, among women who reported contraceptive use 4 months postabortion. Immediate postabortion contraceptive initiators selected a method (injectables, pills or condoms) at the health facility immediately following abortion, and delayed initiators did not select a method immediately following abortion, indicating contraceptive initiation after leaving the health facility.
This study assessed all potential correlates on the day of the abortion procedure. At the individual level, we considered history of MR, contraceptive use at the index pregnancy (resulting in abortion) and fertility intentions for the index pregnancy. Contraceptive use at the index pregnancy included women who reported pregnancy resulting from contraceptive failure or inconsistent use. To measure intentions, interviewers asked women, “Right before you became pregnant, did you want to become pregnant then, did you want to wait until later, did you not want to have any (more) children, or did you not think about it?” [
]. We ordered the husband/partner's intentions from highest to lowest desire for fertility and created three categories relative to the woman's intentions: concordant, discordant–higher (indicating the husband had higher desire for fertility) and discordant–lower (indicating the husband had lower desire for fertility). We excluded discordant–lower from analyses due to the small number (n=7). We included three domains of women's power within the family: past year IPV, accompaniment to the health facility and women's involvement in household decision-making. Service delivery characteristics included abortion treatment and procedure type and postabortion contraceptive counseling. Interviewers asked women to report their abortion treatment type as MR, PAC for induced abortion or PAC for miscarriage.
2.3 Data analysis
We calculated the prevalence of the primary outcome, contraceptive use 4 months postabortion, for the sample and based on potential correlates in the three levels (individual, family and abortion service delivery). F-tests from simple logistic regression models assessed bivariate associations. We ran separate multivariable logistic regression models for each level, including potential correlates with bivariate significance at p<.05. Model 1 was the individual level model; Model 2 was the abortion service delivery level model; and Model 3 was a full model with potential correlates from both levels. All models adjusted for sociodemographic characteristics associated with the outcome (education, number of children and cohabitation with the husband/partner). We used a similar approach for the secondary outcome, timing of postabortion contraceptive initiation. We calculated the prevalence for the sample and by potential correlates in each level and assessed bivariate differences using an F-test. We conducted post hoc analysis of past year IPV experience by timing of initiation and stratified by accompaniment to the health facility to clarify whether the association varied by spousal accompaniment. Approximately 8% of observations were missing for past year IPV, and we generated 10 imputations with multivariate imputation using chained equations [
]. We analyzed the multiple imputation dataset using Stata/SE 14.0, accounting for the complex survey design.
2.4 Ethical review
Study procedures received ethical approval from the Bangladesh Medical Research Council in Dhaka and the Allendale Investigational Review Board in the United States.
Interviewers approached a total of 555 women for participation in the parent study, enrolled 498 women (response rate: 498/555, 90%) and retained 457 women at the 4-month follow-up (follow-up rate: 457/498, 92%). The analytic sample for this study included 398 women who completed the follow-up survey and did not intend pregnancy soon (Fig. 1). Loss to follow-up was nondifferential by all sociodemographic characteristics except parity; loss to follow-up was more common among nulliparous women (16%), compared to parous women (6%; p<.01).
On average, women in the sample were 28 years old, more than half (56.0%) had secondary or higher education and 86.9% had at least one child (Table 1). At the 4-month follow-up 85.4% of women reported contraceptive use. Contraceptive use 4 months postabortion was associated with having at least one child (88.9% among women with 1–2 children and 83.8% among women with 3 or more children vs. 73.1% among women with no children; p<.01) and cohabitating with the husband/partner (86.7% vs. 67.0%; p=.01).
Table 1Sociodemographic characteristics of study sample by contraceptive use 4 months postabortion (n=398)
Total (n = 398)
Not using (n=58)
Contraceptive use 4 months postabortion
Age [mean (SE)]
Husband/Partner's age [mean (SE)]
Secondary or higher
Secondary or higher
Number of children
3 or more children
Cohabitation with husband/partner
Rural to urban migrant
Note: Original n, imputed percent and F-test p-value presented.
Bivariate analysis demonstrated that contraceptive use 4 months postabortion was associated with individual and abortion service delivery characteristics but not family characteristics (Table 2). Ninety-two percent of women who reported contraceptive use at the index pregnancy reported contraceptive use 4 months postabortion, compared to only 76.9% of those who did not report use at the index pregnancy (p<.01). Eighty-eight percent of women who said that the pregnancy was mistimed or unwanted reported contraceptive use 4 months postabortion, compared to 76.3% who said the pregnancy was wanted then or they were ambivalent about the timing (p<.01). Abortion treatment type was also associated; 90.6% of women who received PAC services for induced abortion reported contraceptive use 4 months postabortion, compared to 87.3% of MR clients and 76.4% of PAC for miscarriage clients (p=.03).
Table 2Bivariate associations between potential individual, family and abortion service delivery correlates and contraceptive use 4 months postabortion (n = 398)
In multivariable analyses (Table 3), Model 1 demonstrated that women who reported contraceptive use at the index pregnancy had three times higher odds of reported contraceptive use 4 months postabortion, compared to women who did not report contraceptive use at the index pregnancy (95% confidence interval (CI): 1.5–6.1). Model 2 showed that abortion service delivery characteristics were not associated with contraceptive use 4 months postabortion after adjusting for sociodemographic characteristics. The full model (Model 3) demonstrated that after adjusting for all correlates with bivariate associations with the outcome, only contraceptive use at the index pregnancy was correlated with contraceptive use 4 months postabortion (adjusted odds ratio (AOR)=2.9; 95% CI: 1.5–5.9).
Table 3Logistic regression results of association between contraceptive use 4 months postabortion and potential correlates (n = 398)
Model 1: individual characteristics
Model 2: abortion service delivery characteristics
Timing of postabortion contraceptive initiation was associated with characteristics at the family and abortion service delivery levels (Table 4). At the family level, 44.4% of women whose husband/partner had a higher desire for fertility delayed initiation, compared to only 21.9% of women who reported intentions concordant with their husband/partner's (p=.04). Delayed initiation was also associated with reported past year IPV; 36.8% of women who reported IPV delayed initiation, compared to 19.5% of women who did not report past year IPV (p=.03). At the abortion service delivery level, 56.7% of MA clients and 57.1% of D&C clients delayed initiation, compared to only 12.6% of MVA clients (p<.01).
Table 4Bivariate associations between potential individual, family and abortion service delivery correlates and timing of postabortion contraceptive initiation (n = 340)
We analyzed the interaction between reported past year IPV experience and spousal accompaniment on timing of initiation to explore a possible explanation for delayed initiation. Among women whose husband/partner accompanied them for the abortion procedure, 49.6% of women who reported past year IPV delayed initiation, compared to 19.6% of women who did not report IPV (p=.03) (Table 5). We did not observe significant differences among women who attended the facility alone or were accompanied by someone other than their husband/partner.
Table 5Timing of postabortion contraceptive initiation by past year IPV, stratified by accompaniment to the health facility for the abortion procedure (n = 340)
In this prospective study of abortion clients, reported contraceptive use 4 months postabortion was prevalent (85.4%), and contraceptive use at the index pregnancy was the primary correlate. Our analysis of timing of postabortion contraceptive initiation extended past analyses that used a 1- to 3-month initiation window and identified correlates of delayed compared to immediate initiation. Delayed contraceptive initiation was more common among women who received MA or D&C, women who reported discordant pregnancy intentions and women who reported past year IPV, particularly with spousal accompaniment for the abortion procedure. Though these women were equally likely to report contraceptive use 4 months postabortion, findings highlight ways health system interventions can prevent delays in postabortion contraceptive initiation.
The primary correlate of contraceptive use 4 months postabortion was contraceptive use at the index pregnancy. Findings echo results from other settings demonstrating that contraceptive history predicts postabortion contraceptive use [
], and postabortion contraceptive counseling provides an opportunity to assess previous contraceptive use patterns and provide accurate information on using contraception effectively.
Timing of postabortion contraceptive initiation varied by family and abortion service delivery characteristics. At the abortion service delivery level, MA and D&C clients delayed initiation compared to MVA clients, suggesting providers differentially offered postabortion contraception based on procedure type. In practice, providers often ask MA and D&C clients to return for a follow-up visit to ensure that the abortion is complete and that there are no complications. Providers may prefer to provide postabortion contraception during follow-up visits, but women may miss the opportunity for contraceptive initiation before fertility returns.
This study identified discordant fertility intentions and past year IPV as correlates of delayed initiation, particularly with spousal accompaniment for abortion. Though there were no differences in contraceptive use 4 months postabortion, the timing of initiation differed. This suggests that women reporting IPV are equally able to use contraception by 4 months postabortion but are less likely to initiate postabortion contraception on the day of the procedure. Delayed initiation among women reporting IPV is particularly concerning as studies from a variety of settings demonstrate less power within relationships to negotiate contraceptive use [
], which could result in delays beyond the 2-week period of natural protection from pregnancy after an abortion procedure. This study identified spousal accompaniment as a potential explanation for delayed initiation among women reporting past year IPV. While husband/partner accompaniment to abortion services is seen as positive and supportive for some women in the South Asian context [
], it may be experienced as controlling or coercive when violence is present. Abortion service providers should be trained to provide confidential counseling in a private place within the health facility to screen for violence [
], and postabortion contraceptive counseling provides an opportunity for providers to assess women's needs, match postabortion contraceptive recommendations to these needs and provide information on available support services [
The primary limitations of this study were the small sample size and short follow-up period. Similar studies have used 6-month or 1-year follow-up periods, which provide greater power to assess outcomes such as subsequent pregnancy [
]. Despite the small sample size, a major strength of this study was the low rate of loss to follow-up. Loss to follow-up was nondifferential by most sociodemographic characteristics, but we note potential for selection bias in that nulliparous women were more likely to be lost to follow-up. Contraceptive use 4 months postabortion was likely lowest among nulliparous women, and differential loss could overestimate contraceptive use 4 months postabortion. We relied on self-reported contraceptive use, which is subject to social desirability bias. Calendar data on contraceptive use over the follow-up period would provide more nuanced information on timing of postabortion contraceptive initiation, including whether women used the contraceptive method selected immediately following abortion. We also lacked data on pregnancy risk after abortion, such as sexual activity and resumption of menses. The parent study focused on government and NGO facilities receiving an intervention to improve abortion services; the extent to which findings generalize to other settings, women under age 18 and women who selected long-acting or permanent methods immediately following abortion is unclear.
Contraceptive use at the index pregnancy was the primary correlate of postabortion contraceptive use 4 months postabortion. Postabortion contraceptive counseling should assess contraceptive failure and inconsistent contraceptive use and empower women with accurate information to use contraception effectively. This study also found that women “catch up” in terms of contraceptive use over the 4 months following abortion, but delayed initiation may indicate a need for postabortion contraception not met immediately following abortion, potentially putting women at risk of unwanted pregnancy. This gap in contraceptive coverage is particularly concerning for women reporting IPV, who were more likely to delay initiation, especially if accompanied by a husband/partner for abortion. Interventions should improve confidential counseling to screen for violence and appropriately match postabortion contraceptive provision with women's needs.
We are grateful to our funding sources, including Ipas, the National Institute of Child Health and Human Development T32 predoctoral traineeship on preventing and addressing violence in families and the Sommer Scholars program at the Johns Hopkins Bloomberg School of Public Health. We are also grateful to our partners at the Directorate General of Family Planning and Directorate General of Health Services for supporting this study and to Altaf Hussain and the Bangladesh Association for Prevention of Septic Abortion for collecting study data.
World Health Organization (WHO)
Safe abortion: technical and policy guidance for health systems.