- Email: [email protected]
Contraceptive discontinuation and pregnancy postabortion in Nepal: a longitudinal cohort study
Contraception 91 (2015) 301 – 307
Original research article
Contraceptive discontinuation and pregnancy postabortion in Nepal: a longitudinal cohort study☆,☆☆ Mahesh Puri a , Jillian T. Henderson b, c , Cynthia C. Harper b , Maya Blum b , Deepak Joshi a , Corinne H. Rocca b,⁎ b
a Center for Research on Environment Health & Population Activities (CREHPA), Kusunti, Lalitpur, PO Box 9626, Kathmandu, Nepal Bixby Center for Global Reproductive Health, Department of Obstetrics, Gynecology and Reproductive Sciences, School of Medicine, University of California, 3333 California Street, Suite 335, Box 0744, San Francisco, CA 94143-0744, USA c Kaiser Permanente Center for Health Research, Northwest, 3800 N. Interstate Avenue, Portland, OR 97227, USA Received 14 May 2014; revised 17 December 2014; accepted 22 December 2014
Abstract Objective: To examine postabortion contraceptive discontinuation and pregnancy in Nepal, where abortion was decriminalized in 2002. Study Design: We conducted an observational cohort study of 654 women obtaining abortions from four public and nongovernmental facilities in 2011. Patients completed questionnaires at their abortion visit and 6 and 12 months later. We used Cox proportional hazards models to assess contraceptive discontinuation and pregnancy by method initiated postabortion and other sociodemographic and reproductive factors. Results: Among the 78% (508/654) of women who initiated a modern contraceptive method within 3 months postabortion, the 1-year contraceptive discontinuation rate was 62 per 100 person-years. Discontinuation was far lower among the 5% of women using long-acting reversible methods (21/ 100 person-years) than among those using condoms (74/100 person-years), pills (61/100 person-years) and the injectable [64/100 person-years; adjusted hazard ratio (aHR)=0.32 (0.15–0.68)]. Unmarried women and those not living with their husband experienced higher contraceptive discontinuation [aHR=2.16 (1.47–3.17)]. The 1-year pregnancy rate for all women was 9/100 person-years. Pregnancy was highest among those who initiated no modern method postabortion (13/100 person-years) and condoms (12/100 person-years), and pregnancy was lowest among users of longacting reversible methods (3/100 person-years). The poorest women were at increased pregnancy risk [aHR=2.31 (1.32–4.10)]. Conclusion: Women using intrauterine devices and implants experienced greatly reduced contraceptive discontinuation and pregnancy within a year postabortion, although initiation of these long-acting methods was low. Increased availability of long-acting methods in Nepal and similar settings may help to prevent unwanted pregnancy and attendant maternal mortality and morbidities. Implications: Initiation of modern contraception was high postabortion; however, 1-year discontinuation was high for the condom, pill and injectable, the methods most commonly used. Rates for intrauterine devices and implants were low. Results support efforts to facilitate patient knowledge and access to the full range of contraceptives, including long-acting reversible methods. © 2015 Elsevier Inc. All rights reserved. Keywords: Postabortion contraception; Continuation; Pregnancy; Long-acting reversible contraception
1. Introduction ☆
Acknowledgement of funding: This research was funded by the Society of Family Planning. The views and opinions expressed are those of the authors and do not necessarily represent the views and opinions of the funding agency. ☆☆ Conflicts of interest: We have no conflicts of interests to disclose. ⁎ Corresponding author. Bixby Center for Global Reproductive Health, Department of Obstetrics, Gynecology and Reproductive Sciences, School of Medicine, University of California, 3333 California Street, Suite 335, Box 0744, San Francisco, CA 94143-0744, USA. Tel.: + 1 415 476 6973; fax: +1 415 502 8479. E-mail address: [email protected] (C.H. Rocca). http://dx.doi.org/10.1016/j.contraception.2014.12.011 0010-7824/© 2015 Elsevier Inc. All rights reserved.
Since abortion was legalized in 2002, Nepal has experienced a rapid expansion of safe services [1]. The country's Safe Abortion Policy includes a mandate that abortion facilities provide comprehensive contraceptive care [2]. Effective contraceptive use, including after abortion, is critical to reducing unintended pregnancy and improving women's reproductive autonomy and health [3]. Contraception is also an important intervention to address maternal mortality, which, despite improvements, remains high in many regions of Nepal [4,5].
302
M. Puri et al. / Contraception 91 (2015) 301–307
After increasing from 26% to 44% from 1996 to 2006, contraceptive prevalence in Nepal has since remained level [6]. Although couples increasingly desire smaller families, women rely primarily on sterilization for contraception; use of modern, nonpermanent methods for birth spacing is low, and only 2% use highly effective, long-acting reversible contraceptives [LARC; intrauterine devices (IUDs) and implants] [6]. As in other low-income settings, discontinuation is high, with over half of couples ceasing use of pills, injectables and condoms within a year [3,6]. Concerns about contraceptive nonuse postabortion and a high return to unwanted fertility have emerged in Nepal [7,8]. However, prospective data on contraceptive use patterns and pregnancy postabortion are scant. National data from public facilities indicate that half of patients in 2009–2011 received a contraceptive method; there are no data on specific methods initiated or from private facilities, which provide a majority of legal abortions [1]. Results from analyses of baseline data from the current study — conducted among 838 women at four major public and private abortion facilities — found that 62% of patients chose an effective contraceptive (hormonal, long-acting or permanent) at the time of abortion, but only 44% received supplies [9]. A study using 2011 Demographic and Health Surveys (DHS) data on women who had had an abortion within 5 years similarly found low contraceptive use in the year postabortion (56%), with almost half discontinuing within a year [10]. With the exception of a study in India [11], little research has prospectively examined contraceptive use, discontinuation and incident pregnancy postabortion in the South Asian context. We examined postabortion contraceptive initiation, discontinuation and pregnancy over a year among women receiving legal abortion services in Nepal. We assessed discontinuation and pregnancy by method initiated and other patient characteristics to identify which women are at highest risk. Our aim was to identify areas for improvement in postabortion contraceptive care. 2. Materials and methods 2.1. Study population and procedures We conducted a longitudinal cohort study of contraceptive use postabortion at four major health facilities in Nepal. The study sites were selected purposively to obtain sufficient numbers of eligible participants and diverse settings. They included two nongovernmental clinics and two government hospitals: the clinics were high-volume reproductive health care facilities located in the Western and Eastern regions. The hospitals, in the capital city Kathmandu and in the Terai region bordering India, were both large referral centers for surrounding districts. Women aged 16–35 years having aspiration or medication abortions were enrolled between March and May 2011. Female research assistants approached women after the aspiration abortion or completing the first medication
abortion visit for mifepristone, explained study procedures, conducted age eligibility screening, and obtained verbal informed consent in a private location in the facility. Participants completed interviewer-administered, structured questionnaires asking about sociodemographic information and contraceptive services received. Contact information and preferences were recorded so participants could be contacted for follow-up. Participants were offered a handout with investigator and local institutional review board contact information. At 6 and 12 months, research assistants contacted participants to complete follow-up interviews, conducted in private locations of the participant's choosing, such as home or a clinic. Interviews asked about the contraceptive methods women had used since enrollment and pregnancies experienced. Participants received small remunerative gifts after each interview. Data were kept confidential, and questionnaires contained no identifying information. Procedures are explained in detail elsewhere [9]. The study protocol received ethics approval from the University of California, San Francisco, Committee on Human Research and the Nepal Health Research Council, Kathmandu. The overall study sample size was 838 women, of whom 654 (78%) completed a follow-up visit and contributed longitudinal data. For this analysis, 505 of these 654 women initiated a reversible modern method, comprising the analysis population for contraceptive continuation. This sample size (n=505) was sufficient to measure differences in 12-month continuation by method: we had over 0.90 power with a two-sided test for survival data with an alpha of .05 to detect a difference by method, with uneven numbers per method (ratio of 1:7 of LARC to injectable and 1:3 LARC to condom and pill). 2.2. Measures 2.2.1. Independent variables The primary independent variable was contraceptive method initiated. At each follow-up interview, women reported contraceptive methods they used in each month since the time of the prior interview. Women reporting using condoms, oral contraceptive pills, injections, LARCs or female or male (husband) sterilization within 3 months of baseline were coded as initiating that modern contraceptive method. Women who initiated more than one method were coded according to the more effective method. Baseline questionnaires captured sociodemographic characteristics, reproductive history and pregnancy attitudes, assessed with items on how happy the participant and her husband would be if she became pregnant within 6 months (very happy/happy vs. unhappy/very unhappy). Education was assessed by asking participants the highest level of schooling they completed (≤primary vs. N primary). Household assets were measured using a standardized scale of amenities in women's households (e.g., radio). Receipt of contraceptive counseling was a dichotomous variable of whether the participant reported that a staff member had discussed any contraceptive method with her at her abortion visit.
M. Puri et al. / Contraception 91 (2015) 301–307
2.2.2. Outcome variables Study outcomes were contraceptive method discontinuation and incident pregnancy. Reports of monthly contraceptive methods used from follow-up questionnaires were used to assess the time until discontinuation among women initiating a method. We also created a categorical variable describing patterns of contraceptive use over time, capturing method switching and gaps in use. Pregnancy and time until pregnancy were assessed from questionnaires. Participants reported retrospectively whether new pregnancies were wanted at that time, versus wanted but not at that time or not wanted at all. 2.3. Analysis Analyses included participants completing at least one follow-up questionnaire (n= 654). We used life table analysis accounting for censoring to estimate the contraceptive discontinuation rate among women who initiated a contraceptive method within 3 months postabortion. We also calculated rates by method and graphed the Kaplan–Meier function. Each woman contributed observation time to the analysis until she discontinued her method, was lost-tofollow-up or exited the study at 12 months. Cox proportional hazards models for time-to-event data were estimated to assess differences in rates by method initiated and other baseline factors [12]. The multivariable model excluded characteristics found to be strongly correlated with others in the model. We estimated the Schoenfeld residuals, to check the proportional hazards assumptions; we also plotted log– log curves vs. log-time for each category of independent variables to assess violations of proportionality. We reported contraceptive use patterns by method and assessed reasons for method discontinuation. For pregnancy, we used life table analysis to estimate overall and method-specific pregnancy rates. Cox proportional hazards models for time to first pregnancy were estimated to assess associated factors, including contraceptive method and baseline covariables. We repeated pregnancy analyses limiting the outcome to pregnancies reported as unwanted or occurring earlier than wanted. We conducted attrition analyses. All analyses were conducted using Stata 13 (College Station, TX, USA). Results are reported at the pb.05 level. 3. Results At baseline, most participants were married (98%) and parous (91%) (Table 1). About 37% wanted to have (more) children in the future, and 15% reported they would feel happy if they had a child within 12 months. Overall, 78% of the cohort (508/654) initiated a modern contraceptive method within 3 months postabortion, most commonly the injectable (37%; n= 241), condoms (18%; n= 116) or oral contraceptive pills (17%; n= 114) (Table 2). For LARC, 5.2% (n= 34) initiated either an IUD (3.8%; n= 25) or an implant (1.4%; n= 9); 0.5% (n= 3) had female or male
303
Table 1 Participant characteristics at baseline (n= 654) Age, mean, SD Married, n, % Not currently living with husband or partner, n, % More than a primary education, n, % Lowest quartile asset level Rural residence, n, % Caste/Ethnicity Upper caste groups Relatively advantaged indigenous Disadvantaged indigenous Untouchable and religious minority Parity, n, % Nulliparous Parous, 1 child Parous, 2+ children Wants more children, n, % Happy if pregnant in next 6 months, n, % Husband happy if pregnant in next 6 months, n, % Abortion type, n, % Aspiration Medication Received contraceptive counseling, n, % Facility, n, % Nongovernmental clinic Public hospital
26.8 644 85 466 148 552
4.5 98.5 13.0 71.3 22.6 38.5
299 157 100 68
47.9 25.2 16.0 10.9
62 205 387 242 83 113
9.5 31.4 59.2 37.0 12.7 17.3
517 137 458
79.1 21.0 70.0
331 323
50.6 49.4
sterilization. Over 1-year follow-up, the overall discontinuation rate for those initiating a modern method was 62 per 100 person-years (PY). Discontinuation was highest for condoms (74/100 PY), similar for pills (61/100 PY) and injectables (64/100 PY), and lowest for LARC (21/100 PY) (Fig. 1). In adjusted analyses, LARC discontinuation was lower than for the injectable [adjusted hazard ratio (aHR)= 0.32, 95% confidence interval (CI): 0.15–0.68], pills and condoms (Table 3). Unmarried women and those not currently living with a husband had over double the discontinuation rate as women living with husbands (aHR= 2.16, 95% CI: 1.47–3.17), and those who wanted more children at the time of the abortion experienced significantly higher discontinuation (aHR=1.61, 95% CI: 1.23–2.12). Independent variables met proportionality assumptions, except that discontinuation of the injectable was not Table 2 Contraceptive initiation, discontinuation and pregnancy postabortion, by method initiated (n=654)
No modern method Condoms Pills Injectable IUD/implant Female or male sterilization
Initiation
1-Year discontinuation
1-Year pregnancy
n
(%)
n
(Rate per 100 PY)
n
(Rate per 100 PY)
146 116 114 241 34 3
(22.3) (17.7) (17.4) (36.9) (5.2) (0.5)
– 86 70 154 7 0
– (74.1) (61.4) (63.9) (20.6) (0.0)
24 20 16 22 1 0
(13.3) (12.4) (7.2) (6.4) (2.9) (0.0)
M. Puri et al. / Contraception 91 (2015) 301–307
0.00
0.25
0.50
0.75
Probabilty of discontinuation
1.00
304
0
3
6
9
12
Months Condoms Pills a
Injectable IUD/Implant
Excludes the 3 participants who initiated female or male sterilization
Fig. 1. Kaplan–Meier failure estimates for contraception discontinuation over 1 year after abortion, by method initiated (n= 505 a).
proportional to that of the other methods for the first 3 months postabortion. Patterns of contraceptive use over a year were similar between pill and injectable users: discontinuation with no resumption of another effective method was more common than
method switching and gaps in use (Table 4). Among LARC users, there was virtually no method switching or gaps in use. The most frequently reported reason for discontinuation was fear of side effects or health concerns (34%), followed by inconvenient to use (24%), husband or partner is away (17%), infrequent sex (11%) and desire for a more effective method (10%). Side effects or health concerns was the most common reason for discontinuation among women using pills, injectables and LARCs. For condom users, inconvenience, lack of access and husband or partner disapproval were the most comments reasons for discontinuation. Overall, 56 women reported a pregnancy in the year postabortion (rate=9/100 PY). Pregnancy rates by method initiated were 13/100 PY for no method, 12/100 PY for condoms, 7/100 PY for pills, 6/100 PY for injectables and 3/100 PY for LARC users (Table 2). Compared to women using the injectable, those who did not use a method of contraception postabortion (aHR=2.79, 95% CI: 1.32–5.88) and those using condoms (aHR=2.35, 95% CI: 1.10–5.00) experienced far higher pregnancy (Table 5 and Fig. 2). While the pregnancy rate among LARC initiators was lower, there was only 1 pregnancy among them, too few to test statistical differences. Women who reported wanting more children (aHR=3.44, 95% CI: 1.73–6.85) and those of lowest asset quartile (aHR=2.31, 95% CI: 1.31–4.10) experienced higher
Table 3 Postabortion contraceptive discontinuation over 1 year, Cox proportional hazard analysis (n=505 a) Unadjusted models
Method initiated Condoms Pills Injectable IUD/implant Age Not currently living with husband/partner More than a primary education Lowest quartile asset level Rural residence Caste/Ethnicity Upper caste groups Relatively advantaged indigenous Disadvantaged indigenous Untouchable and religious minority Parity, at least 2 living children Wants more children in future Happy if pregnant within 6 months Husband happy if pregnant within 6 months Abortion type Aspiration Medication Received contraceptive counseling Public facility a
Adjusted model
HR
(95% CI)
aHR
(95% CI)
1.42⁎⁎ † 1.10 † Reference 0.27⁎⁎⁎ 0.98 2.27⁎⁎⁎ 1.27 0.96 1.11
(1.09–1.85) (0.83–1.46)
(1.08–1.85) (0.86–1.53)
(0.12–0.57) (0.95–1.00) (1.56–3.31) (0.99–1.63) (0.74–1.25) (0.89–1.39)
1.41⁎ † 1.15 † Reference 0.32⁎⁎ 1.01 2.16⁎⁎⁎
(0.15–0.68) (0.98–1.04) (1.47–3.17)
1.04 1.03
(0.79–1.36) (0.82–1.29)
Reference 0.84 1.04 0.71 0.76⁎ 1.56⁎⁎⁎ 1.60⁎⁎ 1.43⁎⁎
(0.59–1.18) (0.79–1.37) (0.45–1.09) (0.60–0.94) (1.25–1.95) (1.17–2.17) (1.09–1.88)
1.61⁎⁎⁎
(1.23–2.12)
Reference 0.95 1.26 0.65⁎⁎⁎
(0.73–1.25) (0.97–1.63) (0.52–0.81)
Reference 0.90 1.05 0.71⁎⁎
(0.69–1.19) (0.78–1.41) (0.55–0.92)
Excludes the three participants who initiated female or male sterilization. ⁎ p Valueb.05. ⁎⁎ p Valueb.01. ⁎⁎⁎ p Valueb.001. † Differs from IUD/implant at p valueb.001.
M. Puri et al. / Contraception 91 (2015) 301–307
305
Table 4 Contraceptive use patterns over 1 year postabortion, percentages by method initiated (n=480 a) Contraceptive use pattern
Condoms (n= 108)
Pills (n=106)
Injectable (n=232)
IUD/Implant (n=34)
All reversible methods (n= 480)
Same method, continuous use Same method, gaps in use Switched to other effective method, continuous use Switched to other effective method, gaps in use Discontinued use, no other effective method initiated
27.8 19.4 7.4 9.3 36.1
41.5 4.7 2.8 5.7 45.3
37.5 3.5 6.9 4.7 47.4
79.4 0.0 0.0 2.9 17.7
39.2 7.1 5.6 5.8 42.3
a
Excludes the three participants who initiated female or male sterilization and participants not completing the 1-year interview.
rates of pregnancy. Pregnancy hazard was not proportional for LARC users; all other variables met the proportionality assumption. The rate for pregnancies reported retrospectively as being unwanted or occurring too early was 6/100 PY; 63% of all pregnancies fell in this category. Results were similar when examining only pregnancies reported retrospectively as unwanted or mistimed. In attrition analyses, participants contributing data to these analyses were older, were more educated, were more urban, had more assets and were more likely to live without husbands/partners than those lost-to-follow-up after baseline. They were also more likely to have had abortions at nongovernmental clinics.
4. Discussion This longitudinal study examining postabortion contraception and pregnancy in Nepal found high contraceptive uptake, with over three-quarters of patients starting a modern method within 3 months. However, despite the ostensible availability of all reversible methods at study sites, participants relied primarily on the injectable and pills, while only 5% had an IUD or implant placed. Continuation was low for condom, pill and injectable users, while patients initiating LARC experienced high continuation and improved protection from pregnancy over a year. These results among postabortion patients mirror national estimates: most
Table 5 Pregnancy over 1 year postabortion, Cox proportional hazard analysis (n= 651) Unadjusted models
Adjusted model
HR
(95% CI)
aHR
(95% CI)
2.32⁎ 2.20⁎ 1.26 Reference 0.50 0.89⁎⁎⁎ 0.95 1.02 1.96⁎ 1.39
(1.16–4.63) (1.04–4.60) (0.53–3.02)
(1.32–5.88) (1.10–5.00) (0.62–3.60)
(0.07–3.80) (0.83–0.94) (0.43–2.10) (0.57–1.81) (1.14–3.39) (0.82–2.36)
2.79⁎⁎ 2.35⁎ 1.50 Reference 0.57 0.96 0.55
(0.07–4.36) (0.89–1.04) (0.23–1.33)
2.31⁎⁎ 1.54
(1.31–4.10) (0.90–2.63)
Reference 1.51 1.22 1.43 0.23⁎⁎⁎ 4.25⁎⁎⁎ 2.25⁎⁎ 1.56
(0.73–3.12) (0.62–2.38) (0.61–3.34) (0.13–0.41) (2.41–7.52) (1.21–4.19) (0.84–2.89)
3.44⁎⁎⁎
(1.73–6.85)
Reference 0.91 0.78 1.12
(0.47–1.77) (0.45–1.34) (0.66–1.89)
Reference 0.74 0.95 1.13
(0.38–1.46) (0.48–1.89) (0.59–2.16)
a
Method initiated No effective method Condom Pills Injectable IUD/implant Age Not currently living with husband/partner More than a primary education Lowest quartile asset level Rural residence Caste/Ethnicity Upper caste groups Relatively advantaged indigenous Disadvantaged indigenous Untouchable and religious minority Parity, at least 2 living children Wants more children in future Happy if pregnant within 6 months Husband happy if pregnant within 6 months Abortion type Aspiration Medication Received contraceptive counseling Public facility a
Excludes the three participants who initiated female or male sterilization. ⁎ p Valueb.05. ⁎⁎ p Valueb.01. ⁎⁎⁎ p Valueb.001.
M. Puri et al. / Contraception 91 (2015) 301–307
0.20 0.15 0.10 0.05 0.00
Probability of Pregnancy
0.25
306
0
3
6
9
12
Months No method Condom Pill a
Injectable IUD/implant
Excludes the 3 participants who initiated female or male sterilization
Fig. 2. Kaplan–Meier failure estimates for pregnancy over 1 year after abortion, by method initiated (n=651 a.).
Nepali women initiating condoms (63%), pills (71%) and injectables (55%) discontinue within a year [6], as do about half of women reporting initiation of any method within a year postabortion [10]. While national data on LARC discontinuation are not available, DHS from other developing countries have found that 40%–50% of couples discontinue use of condoms and hormonal methods within a year, while only 13% of couples using IUDs do [3]. The World Health Organization considers LARCs to be appropriate for use immediately postabortion and the most effective methods for pregnancy prevention [13]. LARC methods, which can be used without partner awareness and do not require resupply, may be an important option in contexts in which gender norms hinder many women's autonomy and ability to seek services. Awareness of LARC is high among Nepali women (IUD: 83%, implants: 90%), but only 2% use an IUD or implant, a proportion similar to other South Asian countries but lower than most other regions worldwide [6,14]. While counseling is important for choice of effective methods, prior analyses of these data found that many women who decided to use a LARC were not ultimately able to obtain the method [9]. Qualitative data from providers at study sites found that LARC supply shortages and inadequate time, space and trained personnel served as provision barriers [15]. Improved access to LARC in the postabortion setting in Nepal would help women at high risk of unintended pregnancy better achieve their desires to delay or avoid pregnancy. Notably, women who adopted condoms as their primary contraceptive method in this study experienced pregnancy rates on par with women who did not initiate any modern method. The condom discontinuation rate was higher than for other methods, and condom users reported more gaps in use than women using more effective methods. While condoms should be recommended and supplied to women for the purpose of sexually transmitted disease prevention,
counseling should stress that condoms are far less effective than hormonal and LARC methods in preventing pregnancy. Unmarried women and married women not currently living with their husbands, who jointly comprised 13% of the sample, experienced over double the contraceptive discontinuation as their counterparts living with husbands. Prior analyses found these women were already less likely to receive contraceptive counseling and to choose or receive an effective method postabortion [9]. These findings corroborate DHS findings that “husband not living at home” is the most common reason for method discontinuation in Nepal [6], where high proportions of married men migrate [16]. They also support prior findings that married Nepali women with a migrating husband experienced over double the postabortion discontinuation rate as those without a migrating husband [10]. Research into how to address the unique needs of women who have sex infrequently or sporadically, including the potential of pericoital contraceptive methods, is needed [17]. 4.1. Limitations This study was powered to show differences in method continuation but was not sufficiently powered to measure differences in pregnancy by method; however, the rates suggest that true differences are likely. Another limitation is the reliance on women's reports of contraceptive use, which can introduce misreporting or social desirability bias. We used trained, experienced female interviewers, and most participants completed all questionnaires with the same interviewer, improving rapport. Women in analyses were older, more educated, and more likely to live alone than women lost to follow-up after baseline. Given that some of these traits were associated with discontinuation, we may have underestimated continuation in this study. We likely slightly underestimated pregnancy rates as pregnancies occurring in the last weeks of the study may not have detected until after the final study interview. The four recruitment sites were in diverse clinical settings and geographic regions, but are not a representative sample of Nepal abortion facilities, limiting generalizability. Women's contraceptive choices were likely influenced by provider counseling and the availability of supplies at study sites [15,18]; findings may differ at other facilities. 5. Conclusion Nepal has prioritized postabortion contraceptive services in its national abortion policy [2]. Contraceptive initiation was indeed high among abortion patients in this study. However, while continuation was high for the IUD and implant — adopted by only 5% of patients — it was low for the condom, pill and injectable, the most commonly used methods. Our results demonstrate the importance of overall method mix in determining pregnancy rates on a population level, and support ongoing efforts to facilitate patient
M. Puri et al. / Contraception 91 (2015) 301–307
knowledge and access to the full range of contraceptives, including highly effective methods [19]. Preventing unwanted pregnancy among women known to be at risk is essential to reducing attendant morbidity and mortality in countries like Nepal. [6]
Details of ethics approval Human subjects approval for the study protocol was obtained from the University of California, San Francisco, Committee on Human Research (approval date: October 15, 2010; study # 10-03002). The study was also reviewed and approved by the Nepal Health Research Council (approval date: December 20, 2010; reference # 565).
[7]
[8] [9]
[10]
Acknowledgments [11]
We are grateful to Prabhat Lamichhane and Center for Research on Environment Health and Population Activities staff for collecting data, to recruitment facility staff for supporting this research, and to the Nepal Health Research Council for reviewing this study. An earlier version of results was presented at the International Population Conference organized by the International Union for Scientific Study of Population, in Busan, South Korea, August 2013.
[12] [13]
[14]
[15]
References [1] Samandari G, Wolf M, Basnett I, Hyman A, Andersen K. Implementation of legal abortion in Nepal: a model for rapid scaleup of high-quality care. Reprod Health 2012;7:1–1. [2] Ministry of Health and Population, Department of Health Services, Family Health Division. National Safe Abortion Policy, Final Draft. Kathmandu: Ministry of Health and Population; 2003. [3] Ali MM, Cleland JG, Shah IH. Causes and consequences of contraceptive discontinuation: evidence from 60 Demographic and Health Surveys; 2012 [Available from: http://www.who.int/reproductivehealth/ publications/family_planning/9789241504058/en/]. [4] Hussein J, Bell J, Iang MD, Mesko N, Amery J, Graham W. An appraisal of the maternal mortality decline in Nepal. PLoS One 2011;6:e19898. [5] Pant PD, Suvedi BK, Pradhan A, Hulton L, Matthews Z, Maskey M. Improvements in maternal health in Nepal: further analysis of the 2006
[16]
[17]
[18]
[19]
307
Nepal Demographic and Health Survey; 2008 [Available from: http:// www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1& ved=0CCsQFjAA&url=http%3A%2F%2Fwiki.nyayahealth.org%2Ff %2FFA54_maternal_mortality%2B-%2BJuly%2B10-2008. pdf&ei=ZF5xU5eVNovdoATgg4CADQ&usg=AFQjCNErbJSfSrgB CzYZ50ZzSyg0l-Fa-Q&sig2=qHotP-cd5iEIF98R7XD_Dw&bvm= bv.66330100,d.cGU]. Ministry of Health and Population, New ERA, Macro International Inc. Nepal Demographic and Health Survey, 2011. Kathmandu: Ministry of Health and Population; 2012. Moller A, Ofverstedt S, Siwe K. Proud, not yet satisfied: the experiences of abortion service providers in the Kathmandu Valley, Nepal. Sex Reprod Healthc 2012;3:135–40. Thapa S, Neupane S. Risk factors for repeat abortion in Nepal. Int J Gynecol Obstet 2013;120:32–6. Rocca CH, Puri M, Harper CC, Blum M, Dulal B, Henderson JT. Postabortion contraception a decade after abortion legalization. Int J Gynecol Obstet 2014;126:170–4. Padmadas SS, Lyons-Amos M, Thapa S. Contraceptive behavior among women after abortion in Nepal. Int J Gynecol Obstet 2014;127:132–7. Kalyanwala S, Acharya R, Zavier AJF. Adoption and continuation of contraception following medical or surgical abortion in Bihar and Jharkhand, India. Int J Gynecol Obstet 2012;118:S47–51. Cox DR. Regression models and life-tables. J R Stat Soc 1972;B 34:187–220. World Health Organization. Medical eligibility criteria for contraceptive use; 2010 [Available from: http://www.who.int/reproductivehealth/ publications/family_planning/9789241563888/en/]. United Nations. World contraceptive use, 2011; 2011 [Available from: http://www.un.org/esa/population/publications/contraceptive2011/ contraceptive2011.htm]. Wang L, Puri M, Joshi D, Blum M, Henderson J. Post-abortion contraceptive services in Nepal: perspectives of abortion care stakeholders. Annual Meetings of the National Abortion Federation. New York, NY, April 27-30, 2013; 2013. Ban B, Karki S, Shrestha A, Hodgins S. Spousal separation and interpretation of contraceptive use and unmet need in rural Nepal. Int Perspect Sex Reprod Health 2012;38:43–7. Cover JK, Drake JK, Kyamwanga IT, Turyakira E, Dargan T, Kumakech E, et al. Consumer perspectives on a pericoital contraceptive pill in India and Uganda. Int Perspect Sex Reprod Health 2013;39:195–204. Tripney J, Kwan I, Bird K. Postabortion family planning counseling and services for women in low-income countries: a systematic review. Contraception 2013;87:17–25. Sapkota S, Coulson J, Dangol A. Improving the uptake of LARC among clients presenting with post-abortion complications: mobilizing the LAM Champion group in Nepal. International Conference on Family Planning. Addis Ababa, November 12-15, 2013; 2013.
Original research article
Contraceptive discontinuation and pregnancy postabortion in Nepal: a longitudinal cohort study☆,☆☆ Mahesh Puri a , Jillian T. Henderson b, c , Cynthia C. Harper b , Maya Blum b , Deepak Joshi a , Corinne H. Rocca b,⁎ b
a Center for Research on Environment Health & Population Activities (CREHPA), Kusunti, Lalitpur, PO Box 9626, Kathmandu, Nepal Bixby Center for Global Reproductive Health, Department of Obstetrics, Gynecology and Reproductive Sciences, School of Medicine, University of California, 3333 California Street, Suite 335, Box 0744, San Francisco, CA 94143-0744, USA c Kaiser Permanente Center for Health Research, Northwest, 3800 N. Interstate Avenue, Portland, OR 97227, USA Received 14 May 2014; revised 17 December 2014; accepted 22 December 2014
Abstract Objective: To examine postabortion contraceptive discontinuation and pregnancy in Nepal, where abortion was decriminalized in 2002. Study Design: We conducted an observational cohort study of 654 women obtaining abortions from four public and nongovernmental facilities in 2011. Patients completed questionnaires at their abortion visit and 6 and 12 months later. We used Cox proportional hazards models to assess contraceptive discontinuation and pregnancy by method initiated postabortion and other sociodemographic and reproductive factors. Results: Among the 78% (508/654) of women who initiated a modern contraceptive method within 3 months postabortion, the 1-year contraceptive discontinuation rate was 62 per 100 person-years. Discontinuation was far lower among the 5% of women using long-acting reversible methods (21/ 100 person-years) than among those using condoms (74/100 person-years), pills (61/100 person-years) and the injectable [64/100 person-years; adjusted hazard ratio (aHR)=0.32 (0.15–0.68)]. Unmarried women and those not living with their husband experienced higher contraceptive discontinuation [aHR=2.16 (1.47–3.17)]. The 1-year pregnancy rate for all women was 9/100 person-years. Pregnancy was highest among those who initiated no modern method postabortion (13/100 person-years) and condoms (12/100 person-years), and pregnancy was lowest among users of longacting reversible methods (3/100 person-years). The poorest women were at increased pregnancy risk [aHR=2.31 (1.32–4.10)]. Conclusion: Women using intrauterine devices and implants experienced greatly reduced contraceptive discontinuation and pregnancy within a year postabortion, although initiation of these long-acting methods was low. Increased availability of long-acting methods in Nepal and similar settings may help to prevent unwanted pregnancy and attendant maternal mortality and morbidities. Implications: Initiation of modern contraception was high postabortion; however, 1-year discontinuation was high for the condom, pill and injectable, the methods most commonly used. Rates for intrauterine devices and implants were low. Results support efforts to facilitate patient knowledge and access to the full range of contraceptives, including long-acting reversible methods. © 2015 Elsevier Inc. All rights reserved. Keywords: Postabortion contraception; Continuation; Pregnancy; Long-acting reversible contraception
1. Introduction ☆
Acknowledgement of funding: This research was funded by the Society of Family Planning. The views and opinions expressed are those of the authors and do not necessarily represent the views and opinions of the funding agency. ☆☆ Conflicts of interest: We have no conflicts of interests to disclose. ⁎ Corresponding author. Bixby Center for Global Reproductive Health, Department of Obstetrics, Gynecology and Reproductive Sciences, School of Medicine, University of California, 3333 California Street, Suite 335, Box 0744, San Francisco, CA 94143-0744, USA. Tel.: + 1 415 476 6973; fax: +1 415 502 8479. E-mail address: [email protected] (C.H. Rocca). http://dx.doi.org/10.1016/j.contraception.2014.12.011 0010-7824/© 2015 Elsevier Inc. All rights reserved.
Since abortion was legalized in 2002, Nepal has experienced a rapid expansion of safe services [1]. The country's Safe Abortion Policy includes a mandate that abortion facilities provide comprehensive contraceptive care [2]. Effective contraceptive use, including after abortion, is critical to reducing unintended pregnancy and improving women's reproductive autonomy and health [3]. Contraception is also an important intervention to address maternal mortality, which, despite improvements, remains high in many regions of Nepal [4,5].
302
M. Puri et al. / Contraception 91 (2015) 301–307
After increasing from 26% to 44% from 1996 to 2006, contraceptive prevalence in Nepal has since remained level [6]. Although couples increasingly desire smaller families, women rely primarily on sterilization for contraception; use of modern, nonpermanent methods for birth spacing is low, and only 2% use highly effective, long-acting reversible contraceptives [LARC; intrauterine devices (IUDs) and implants] [6]. As in other low-income settings, discontinuation is high, with over half of couples ceasing use of pills, injectables and condoms within a year [3,6]. Concerns about contraceptive nonuse postabortion and a high return to unwanted fertility have emerged in Nepal [7,8]. However, prospective data on contraceptive use patterns and pregnancy postabortion are scant. National data from public facilities indicate that half of patients in 2009–2011 received a contraceptive method; there are no data on specific methods initiated or from private facilities, which provide a majority of legal abortions [1]. Results from analyses of baseline data from the current study — conducted among 838 women at four major public and private abortion facilities — found that 62% of patients chose an effective contraceptive (hormonal, long-acting or permanent) at the time of abortion, but only 44% received supplies [9]. A study using 2011 Demographic and Health Surveys (DHS) data on women who had had an abortion within 5 years similarly found low contraceptive use in the year postabortion (56%), with almost half discontinuing within a year [10]. With the exception of a study in India [11], little research has prospectively examined contraceptive use, discontinuation and incident pregnancy postabortion in the South Asian context. We examined postabortion contraceptive initiation, discontinuation and pregnancy over a year among women receiving legal abortion services in Nepal. We assessed discontinuation and pregnancy by method initiated and other patient characteristics to identify which women are at highest risk. Our aim was to identify areas for improvement in postabortion contraceptive care. 2. Materials and methods 2.1. Study population and procedures We conducted a longitudinal cohort study of contraceptive use postabortion at four major health facilities in Nepal. The study sites were selected purposively to obtain sufficient numbers of eligible participants and diverse settings. They included two nongovernmental clinics and two government hospitals: the clinics were high-volume reproductive health care facilities located in the Western and Eastern regions. The hospitals, in the capital city Kathmandu and in the Terai region bordering India, were both large referral centers for surrounding districts. Women aged 16–35 years having aspiration or medication abortions were enrolled between March and May 2011. Female research assistants approached women after the aspiration abortion or completing the first medication
abortion visit for mifepristone, explained study procedures, conducted age eligibility screening, and obtained verbal informed consent in a private location in the facility. Participants completed interviewer-administered, structured questionnaires asking about sociodemographic information and contraceptive services received. Contact information and preferences were recorded so participants could be contacted for follow-up. Participants were offered a handout with investigator and local institutional review board contact information. At 6 and 12 months, research assistants contacted participants to complete follow-up interviews, conducted in private locations of the participant's choosing, such as home or a clinic. Interviews asked about the contraceptive methods women had used since enrollment and pregnancies experienced. Participants received small remunerative gifts after each interview. Data were kept confidential, and questionnaires contained no identifying information. Procedures are explained in detail elsewhere [9]. The study protocol received ethics approval from the University of California, San Francisco, Committee on Human Research and the Nepal Health Research Council, Kathmandu. The overall study sample size was 838 women, of whom 654 (78%) completed a follow-up visit and contributed longitudinal data. For this analysis, 505 of these 654 women initiated a reversible modern method, comprising the analysis population for contraceptive continuation. This sample size (n=505) was sufficient to measure differences in 12-month continuation by method: we had over 0.90 power with a two-sided test for survival data with an alpha of .05 to detect a difference by method, with uneven numbers per method (ratio of 1:7 of LARC to injectable and 1:3 LARC to condom and pill). 2.2. Measures 2.2.1. Independent variables The primary independent variable was contraceptive method initiated. At each follow-up interview, women reported contraceptive methods they used in each month since the time of the prior interview. Women reporting using condoms, oral contraceptive pills, injections, LARCs or female or male (husband) sterilization within 3 months of baseline were coded as initiating that modern contraceptive method. Women who initiated more than one method were coded according to the more effective method. Baseline questionnaires captured sociodemographic characteristics, reproductive history and pregnancy attitudes, assessed with items on how happy the participant and her husband would be if she became pregnant within 6 months (very happy/happy vs. unhappy/very unhappy). Education was assessed by asking participants the highest level of schooling they completed (≤primary vs. N primary). Household assets were measured using a standardized scale of amenities in women's households (e.g., radio). Receipt of contraceptive counseling was a dichotomous variable of whether the participant reported that a staff member had discussed any contraceptive method with her at her abortion visit.
M. Puri et al. / Contraception 91 (2015) 301–307
2.2.2. Outcome variables Study outcomes were contraceptive method discontinuation and incident pregnancy. Reports of monthly contraceptive methods used from follow-up questionnaires were used to assess the time until discontinuation among women initiating a method. We also created a categorical variable describing patterns of contraceptive use over time, capturing method switching and gaps in use. Pregnancy and time until pregnancy were assessed from questionnaires. Participants reported retrospectively whether new pregnancies were wanted at that time, versus wanted but not at that time or not wanted at all. 2.3. Analysis Analyses included participants completing at least one follow-up questionnaire (n= 654). We used life table analysis accounting for censoring to estimate the contraceptive discontinuation rate among women who initiated a contraceptive method within 3 months postabortion. We also calculated rates by method and graphed the Kaplan–Meier function. Each woman contributed observation time to the analysis until she discontinued her method, was lost-tofollow-up or exited the study at 12 months. Cox proportional hazards models for time-to-event data were estimated to assess differences in rates by method initiated and other baseline factors [12]. The multivariable model excluded characteristics found to be strongly correlated with others in the model. We estimated the Schoenfeld residuals, to check the proportional hazards assumptions; we also plotted log– log curves vs. log-time for each category of independent variables to assess violations of proportionality. We reported contraceptive use patterns by method and assessed reasons for method discontinuation. For pregnancy, we used life table analysis to estimate overall and method-specific pregnancy rates. Cox proportional hazards models for time to first pregnancy were estimated to assess associated factors, including contraceptive method and baseline covariables. We repeated pregnancy analyses limiting the outcome to pregnancies reported as unwanted or occurring earlier than wanted. We conducted attrition analyses. All analyses were conducted using Stata 13 (College Station, TX, USA). Results are reported at the pb.05 level. 3. Results At baseline, most participants were married (98%) and parous (91%) (Table 1). About 37% wanted to have (more) children in the future, and 15% reported they would feel happy if they had a child within 12 months. Overall, 78% of the cohort (508/654) initiated a modern contraceptive method within 3 months postabortion, most commonly the injectable (37%; n= 241), condoms (18%; n= 116) or oral contraceptive pills (17%; n= 114) (Table 2). For LARC, 5.2% (n= 34) initiated either an IUD (3.8%; n= 25) or an implant (1.4%; n= 9); 0.5% (n= 3) had female or male
303
Table 1 Participant characteristics at baseline (n= 654) Age, mean, SD Married, n, % Not currently living with husband or partner, n, % More than a primary education, n, % Lowest quartile asset level Rural residence, n, % Caste/Ethnicity Upper caste groups Relatively advantaged indigenous Disadvantaged indigenous Untouchable and religious minority Parity, n, % Nulliparous Parous, 1 child Parous, 2+ children Wants more children, n, % Happy if pregnant in next 6 months, n, % Husband happy if pregnant in next 6 months, n, % Abortion type, n, % Aspiration Medication Received contraceptive counseling, n, % Facility, n, % Nongovernmental clinic Public hospital
26.8 644 85 466 148 552
4.5 98.5 13.0 71.3 22.6 38.5
299 157 100 68
47.9 25.2 16.0 10.9
62 205 387 242 83 113
9.5 31.4 59.2 37.0 12.7 17.3
517 137 458
79.1 21.0 70.0
331 323
50.6 49.4
sterilization. Over 1-year follow-up, the overall discontinuation rate for those initiating a modern method was 62 per 100 person-years (PY). Discontinuation was highest for condoms (74/100 PY), similar for pills (61/100 PY) and injectables (64/100 PY), and lowest for LARC (21/100 PY) (Fig. 1). In adjusted analyses, LARC discontinuation was lower than for the injectable [adjusted hazard ratio (aHR)= 0.32, 95% confidence interval (CI): 0.15–0.68], pills and condoms (Table 3). Unmarried women and those not currently living with a husband had over double the discontinuation rate as women living with husbands (aHR= 2.16, 95% CI: 1.47–3.17), and those who wanted more children at the time of the abortion experienced significantly higher discontinuation (aHR=1.61, 95% CI: 1.23–2.12). Independent variables met proportionality assumptions, except that discontinuation of the injectable was not Table 2 Contraceptive initiation, discontinuation and pregnancy postabortion, by method initiated (n=654)
No modern method Condoms Pills Injectable IUD/implant Female or male sterilization
Initiation
1-Year discontinuation
1-Year pregnancy
n
(%)
n
(Rate per 100 PY)
n
(Rate per 100 PY)
146 116 114 241 34 3
(22.3) (17.7) (17.4) (36.9) (5.2) (0.5)
– 86 70 154 7 0
– (74.1) (61.4) (63.9) (20.6) (0.0)
24 20 16 22 1 0
(13.3) (12.4) (7.2) (6.4) (2.9) (0.0)
M. Puri et al. / Contraception 91 (2015) 301–307
0.00
0.25
0.50
0.75
Probabilty of discontinuation
1.00
304
0
3
6
9
12
Months Condoms Pills a
Injectable IUD/Implant
Excludes the 3 participants who initiated female or male sterilization
Fig. 1. Kaplan–Meier failure estimates for contraception discontinuation over 1 year after abortion, by method initiated (n= 505 a).
proportional to that of the other methods for the first 3 months postabortion. Patterns of contraceptive use over a year were similar between pill and injectable users: discontinuation with no resumption of another effective method was more common than
method switching and gaps in use (Table 4). Among LARC users, there was virtually no method switching or gaps in use. The most frequently reported reason for discontinuation was fear of side effects or health concerns (34%), followed by inconvenient to use (24%), husband or partner is away (17%), infrequent sex (11%) and desire for a more effective method (10%). Side effects or health concerns was the most common reason for discontinuation among women using pills, injectables and LARCs. For condom users, inconvenience, lack of access and husband or partner disapproval were the most comments reasons for discontinuation. Overall, 56 women reported a pregnancy in the year postabortion (rate=9/100 PY). Pregnancy rates by method initiated were 13/100 PY for no method, 12/100 PY for condoms, 7/100 PY for pills, 6/100 PY for injectables and 3/100 PY for LARC users (Table 2). Compared to women using the injectable, those who did not use a method of contraception postabortion (aHR=2.79, 95% CI: 1.32–5.88) and those using condoms (aHR=2.35, 95% CI: 1.10–5.00) experienced far higher pregnancy (Table 5 and Fig. 2). While the pregnancy rate among LARC initiators was lower, there was only 1 pregnancy among them, too few to test statistical differences. Women who reported wanting more children (aHR=3.44, 95% CI: 1.73–6.85) and those of lowest asset quartile (aHR=2.31, 95% CI: 1.31–4.10) experienced higher
Table 3 Postabortion contraceptive discontinuation over 1 year, Cox proportional hazard analysis (n=505 a) Unadjusted models
Method initiated Condoms Pills Injectable IUD/implant Age Not currently living with husband/partner More than a primary education Lowest quartile asset level Rural residence Caste/Ethnicity Upper caste groups Relatively advantaged indigenous Disadvantaged indigenous Untouchable and religious minority Parity, at least 2 living children Wants more children in future Happy if pregnant within 6 months Husband happy if pregnant within 6 months Abortion type Aspiration Medication Received contraceptive counseling Public facility a
Adjusted model
HR
(95% CI)
aHR
(95% CI)
1.42⁎⁎ † 1.10 † Reference 0.27⁎⁎⁎ 0.98 2.27⁎⁎⁎ 1.27 0.96 1.11
(1.09–1.85) (0.83–1.46)
(1.08–1.85) (0.86–1.53)
(0.12–0.57) (0.95–1.00) (1.56–3.31) (0.99–1.63) (0.74–1.25) (0.89–1.39)
1.41⁎ † 1.15 † Reference 0.32⁎⁎ 1.01 2.16⁎⁎⁎
(0.15–0.68) (0.98–1.04) (1.47–3.17)
1.04 1.03
(0.79–1.36) (0.82–1.29)
Reference 0.84 1.04 0.71 0.76⁎ 1.56⁎⁎⁎ 1.60⁎⁎ 1.43⁎⁎
(0.59–1.18) (0.79–1.37) (0.45–1.09) (0.60–0.94) (1.25–1.95) (1.17–2.17) (1.09–1.88)
1.61⁎⁎⁎
(1.23–2.12)
Reference 0.95 1.26 0.65⁎⁎⁎
(0.73–1.25) (0.97–1.63) (0.52–0.81)
Reference 0.90 1.05 0.71⁎⁎
(0.69–1.19) (0.78–1.41) (0.55–0.92)
Excludes the three participants who initiated female or male sterilization. ⁎ p Valueb.05. ⁎⁎ p Valueb.01. ⁎⁎⁎ p Valueb.001. † Differs from IUD/implant at p valueb.001.
M. Puri et al. / Contraception 91 (2015) 301–307
305
Table 4 Contraceptive use patterns over 1 year postabortion, percentages by method initiated (n=480 a) Contraceptive use pattern
Condoms (n= 108)
Pills (n=106)
Injectable (n=232)
IUD/Implant (n=34)
All reversible methods (n= 480)
Same method, continuous use Same method, gaps in use Switched to other effective method, continuous use Switched to other effective method, gaps in use Discontinued use, no other effective method initiated
27.8 19.4 7.4 9.3 36.1
41.5 4.7 2.8 5.7 45.3
37.5 3.5 6.9 4.7 47.4
79.4 0.0 0.0 2.9 17.7
39.2 7.1 5.6 5.8 42.3
a
Excludes the three participants who initiated female or male sterilization and participants not completing the 1-year interview.
rates of pregnancy. Pregnancy hazard was not proportional for LARC users; all other variables met the proportionality assumption. The rate for pregnancies reported retrospectively as being unwanted or occurring too early was 6/100 PY; 63% of all pregnancies fell in this category. Results were similar when examining only pregnancies reported retrospectively as unwanted or mistimed. In attrition analyses, participants contributing data to these analyses were older, were more educated, were more urban, had more assets and were more likely to live without husbands/partners than those lost-to-follow-up after baseline. They were also more likely to have had abortions at nongovernmental clinics.
4. Discussion This longitudinal study examining postabortion contraception and pregnancy in Nepal found high contraceptive uptake, with over three-quarters of patients starting a modern method within 3 months. However, despite the ostensible availability of all reversible methods at study sites, participants relied primarily on the injectable and pills, while only 5% had an IUD or implant placed. Continuation was low for condom, pill and injectable users, while patients initiating LARC experienced high continuation and improved protection from pregnancy over a year. These results among postabortion patients mirror national estimates: most
Table 5 Pregnancy over 1 year postabortion, Cox proportional hazard analysis (n= 651) Unadjusted models
Adjusted model
HR
(95% CI)
aHR
(95% CI)
2.32⁎ 2.20⁎ 1.26 Reference 0.50 0.89⁎⁎⁎ 0.95 1.02 1.96⁎ 1.39
(1.16–4.63) (1.04–4.60) (0.53–3.02)
(1.32–5.88) (1.10–5.00) (0.62–3.60)
(0.07–3.80) (0.83–0.94) (0.43–2.10) (0.57–1.81) (1.14–3.39) (0.82–2.36)
2.79⁎⁎ 2.35⁎ 1.50 Reference 0.57 0.96 0.55
(0.07–4.36) (0.89–1.04) (0.23–1.33)
2.31⁎⁎ 1.54
(1.31–4.10) (0.90–2.63)
Reference 1.51 1.22 1.43 0.23⁎⁎⁎ 4.25⁎⁎⁎ 2.25⁎⁎ 1.56
(0.73–3.12) (0.62–2.38) (0.61–3.34) (0.13–0.41) (2.41–7.52) (1.21–4.19) (0.84–2.89)
3.44⁎⁎⁎
(1.73–6.85)
Reference 0.91 0.78 1.12
(0.47–1.77) (0.45–1.34) (0.66–1.89)
Reference 0.74 0.95 1.13
(0.38–1.46) (0.48–1.89) (0.59–2.16)
a
Method initiated No effective method Condom Pills Injectable IUD/implant Age Not currently living with husband/partner More than a primary education Lowest quartile asset level Rural residence Caste/Ethnicity Upper caste groups Relatively advantaged indigenous Disadvantaged indigenous Untouchable and religious minority Parity, at least 2 living children Wants more children in future Happy if pregnant within 6 months Husband happy if pregnant within 6 months Abortion type Aspiration Medication Received contraceptive counseling Public facility a
Excludes the three participants who initiated female or male sterilization. ⁎ p Valueb.05. ⁎⁎ p Valueb.01. ⁎⁎⁎ p Valueb.001.
M. Puri et al. / Contraception 91 (2015) 301–307
0.20 0.15 0.10 0.05 0.00
Probability of Pregnancy
0.25
306
0
3
6
9
12
Months No method Condom Pill a
Injectable IUD/implant
Excludes the 3 participants who initiated female or male sterilization
Fig. 2. Kaplan–Meier failure estimates for pregnancy over 1 year after abortion, by method initiated (n=651 a.).
Nepali women initiating condoms (63%), pills (71%) and injectables (55%) discontinue within a year [6], as do about half of women reporting initiation of any method within a year postabortion [10]. While national data on LARC discontinuation are not available, DHS from other developing countries have found that 40%–50% of couples discontinue use of condoms and hormonal methods within a year, while only 13% of couples using IUDs do [3]. The World Health Organization considers LARCs to be appropriate for use immediately postabortion and the most effective methods for pregnancy prevention [13]. LARC methods, which can be used without partner awareness and do not require resupply, may be an important option in contexts in which gender norms hinder many women's autonomy and ability to seek services. Awareness of LARC is high among Nepali women (IUD: 83%, implants: 90%), but only 2% use an IUD or implant, a proportion similar to other South Asian countries but lower than most other regions worldwide [6,14]. While counseling is important for choice of effective methods, prior analyses of these data found that many women who decided to use a LARC were not ultimately able to obtain the method [9]. Qualitative data from providers at study sites found that LARC supply shortages and inadequate time, space and trained personnel served as provision barriers [15]. Improved access to LARC in the postabortion setting in Nepal would help women at high risk of unintended pregnancy better achieve their desires to delay or avoid pregnancy. Notably, women who adopted condoms as their primary contraceptive method in this study experienced pregnancy rates on par with women who did not initiate any modern method. The condom discontinuation rate was higher than for other methods, and condom users reported more gaps in use than women using more effective methods. While condoms should be recommended and supplied to women for the purpose of sexually transmitted disease prevention,
counseling should stress that condoms are far less effective than hormonal and LARC methods in preventing pregnancy. Unmarried women and married women not currently living with their husbands, who jointly comprised 13% of the sample, experienced over double the contraceptive discontinuation as their counterparts living with husbands. Prior analyses found these women were already less likely to receive contraceptive counseling and to choose or receive an effective method postabortion [9]. These findings corroborate DHS findings that “husband not living at home” is the most common reason for method discontinuation in Nepal [6], where high proportions of married men migrate [16]. They also support prior findings that married Nepali women with a migrating husband experienced over double the postabortion discontinuation rate as those without a migrating husband [10]. Research into how to address the unique needs of women who have sex infrequently or sporadically, including the potential of pericoital contraceptive methods, is needed [17]. 4.1. Limitations This study was powered to show differences in method continuation but was not sufficiently powered to measure differences in pregnancy by method; however, the rates suggest that true differences are likely. Another limitation is the reliance on women's reports of contraceptive use, which can introduce misreporting or social desirability bias. We used trained, experienced female interviewers, and most participants completed all questionnaires with the same interviewer, improving rapport. Women in analyses were older, more educated, and more likely to live alone than women lost to follow-up after baseline. Given that some of these traits were associated with discontinuation, we may have underestimated continuation in this study. We likely slightly underestimated pregnancy rates as pregnancies occurring in the last weeks of the study may not have detected until after the final study interview. The four recruitment sites were in diverse clinical settings and geographic regions, but are not a representative sample of Nepal abortion facilities, limiting generalizability. Women's contraceptive choices were likely influenced by provider counseling and the availability of supplies at study sites [15,18]; findings may differ at other facilities. 5. Conclusion Nepal has prioritized postabortion contraceptive services in its national abortion policy [2]. Contraceptive initiation was indeed high among abortion patients in this study. However, while continuation was high for the IUD and implant — adopted by only 5% of patients — it was low for the condom, pill and injectable, the most commonly used methods. Our results demonstrate the importance of overall method mix in determining pregnancy rates on a population level, and support ongoing efforts to facilitate patient
M. Puri et al. / Contraception 91 (2015) 301–307
knowledge and access to the full range of contraceptives, including highly effective methods [19]. Preventing unwanted pregnancy among women known to be at risk is essential to reducing attendant morbidity and mortality in countries like Nepal. [6]
Details of ethics approval Human subjects approval for the study protocol was obtained from the University of California, San Francisco, Committee on Human Research (approval date: October 15, 2010; study # 10-03002). The study was also reviewed and approved by the Nepal Health Research Council (approval date: December 20, 2010; reference # 565).
[7]
[8] [9]
[10]
Acknowledgments [11]
We are grateful to Prabhat Lamichhane and Center for Research on Environment Health and Population Activities staff for collecting data, to recruitment facility staff for supporting this research, and to the Nepal Health Research Council for reviewing this study. An earlier version of results was presented at the International Population Conference organized by the International Union for Scientific Study of Population, in Busan, South Korea, August 2013.
[12] [13]
[14]
[15]
References [1] Samandari G, Wolf M, Basnett I, Hyman A, Andersen K. Implementation of legal abortion in Nepal: a model for rapid scaleup of high-quality care. Reprod Health 2012;7:1–1. [2] Ministry of Health and Population, Department of Health Services, Family Health Division. National Safe Abortion Policy, Final Draft. Kathmandu: Ministry of Health and Population; 2003. [3] Ali MM, Cleland JG, Shah IH. Causes and consequences of contraceptive discontinuation: evidence from 60 Demographic and Health Surveys; 2012 [Available from: http://www.who.int/reproductivehealth/ publications/family_planning/9789241504058/en/]. [4] Hussein J, Bell J, Iang MD, Mesko N, Amery J, Graham W. An appraisal of the maternal mortality decline in Nepal. PLoS One 2011;6:e19898. [5] Pant PD, Suvedi BK, Pradhan A, Hulton L, Matthews Z, Maskey M. Improvements in maternal health in Nepal: further analysis of the 2006
[16]
[17]
[18]
[19]
307
Nepal Demographic and Health Survey; 2008 [Available from: http:// www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=1& ved=0CCsQFjAA&url=http%3A%2F%2Fwiki.nyayahealth.org%2Ff %2FFA54_maternal_mortality%2B-%2BJuly%2B10-2008. pdf&ei=ZF5xU5eVNovdoATgg4CADQ&usg=AFQjCNErbJSfSrgB CzYZ50ZzSyg0l-Fa-Q&sig2=qHotP-cd5iEIF98R7XD_Dw&bvm= bv.66330100,d.cGU]. Ministry of Health and Population, New ERA, Macro International Inc. Nepal Demographic and Health Survey, 2011. Kathmandu: Ministry of Health and Population; 2012. Moller A, Ofverstedt S, Siwe K. Proud, not yet satisfied: the experiences of abortion service providers in the Kathmandu Valley, Nepal. Sex Reprod Healthc 2012;3:135–40. Thapa S, Neupane S. Risk factors for repeat abortion in Nepal. Int J Gynecol Obstet 2013;120:32–6. Rocca CH, Puri M, Harper CC, Blum M, Dulal B, Henderson JT. Postabortion contraception a decade after abortion legalization. Int J Gynecol Obstet 2014;126:170–4. Padmadas SS, Lyons-Amos M, Thapa S. Contraceptive behavior among women after abortion in Nepal. Int J Gynecol Obstet 2014;127:132–7. Kalyanwala S, Acharya R, Zavier AJF. Adoption and continuation of contraception following medical or surgical abortion in Bihar and Jharkhand, India. Int J Gynecol Obstet 2012;118:S47–51. Cox DR. Regression models and life-tables. J R Stat Soc 1972;B 34:187–220. World Health Organization. Medical eligibility criteria for contraceptive use; 2010 [Available from: http://www.who.int/reproductivehealth/ publications/family_planning/9789241563888/en/]. United Nations. World contraceptive use, 2011; 2011 [Available from: http://www.un.org/esa/population/publications/contraceptive2011/ contraceptive2011.htm]. Wang L, Puri M, Joshi D, Blum M, Henderson J. Post-abortion contraceptive services in Nepal: perspectives of abortion care stakeholders. Annual Meetings of the National Abortion Federation. New York, NY, April 27-30, 2013; 2013. Ban B, Karki S, Shrestha A, Hodgins S. Spousal separation and interpretation of contraceptive use and unmet need in rural Nepal. Int Perspect Sex Reprod Health 2012;38:43–7. Cover JK, Drake JK, Kyamwanga IT, Turyakira E, Dargan T, Kumakech E, et al. Consumer perspectives on a pericoital contraceptive pill in India and Uganda. Int Perspect Sex Reprod Health 2013;39:195–204. Tripney J, Kwan I, Bird K. Postabortion family planning counseling and services for women in low-income countries: a systematic review. Contraception 2013;87:17–25. Sapkota S, Coulson J, Dangol A. Improving the uptake of LARC among clients presenting with post-abortion complications: mobilizing the LAM Champion group in Nepal. International Conference on Family Planning. Addis Ababa, November 12-15, 2013; 2013.