Factors associated with unsafe abortion practices in Nepal: Pooled analysis of the 2011 and 2016 Nepal Demographic and Health Surveys.

BACKGROUND: Unsafe abortion contributes to maternal morbidities, mortalities as well as social and financial costs to women, families, and the health system. This study aimed to examine the factors associated with unsafe abortion practices in Nepal.
METHODS: Data were derived from the 2011 and 2016 Nepal Demographic and Health Surveys (NDHS). A total of 911 women aged 15-49 years who aborted five years prior to surveys were included in the analysis. The multivariate logistic regression analysis was employed to determine factors associated with unsafe abortion.
RESULTS: Unsafe abortion rate was seven per 1000 women aged 15-49 years. This research found that women living in the Mountains (adjusted Odds Ratio (aOR) 2.36; 95% CI 1.21, 4.60), or those who were urban residents (aOR 2.11; 95% CI 1.37, 3.24) were more likely to have unsafe abortion. The odds of unsafe abortion were higher amongst women of poor households (aOR 2.16; 95% CI 1.18, 3.94); Dalit women (aOR 1.89; 95% CI 1.02, 3.52), husband with no education background (aOR 2.12; 95%CI 1.06, 4.22), or women who reported agriculture occupation (aOR 1.82; 95% CI 1.16, 2.86) compared to their reference's group. Regardless of knowledge on legal conditions of abortion, the probability of having unsafe abortion was significantly higher (aOR 5.13; 95% CI 2.64, 9.98) amongst women who did not know the location of safe abortion sites. Finally, women who wanted to delay or space childbirth (aOR 2.71; 95% CI 1.39, 5.28) or those who reported unwanted birth (aOR = 2.33; 95% CI 1.19, 4.56) were at higher risk of unsafe abortion.
CONCLUSION: Going forward, increasing the availability of safe abortion facilities and strengthening family planning services can help reduce unsafe abortion in Nepal. These programmatic efforts should be targeted to women of poor households, disadvantaged ethnicities, and those who reside in mountainous region.

Introduction

World Health Organization (WHO) defines unsafe abortion as a procedure for terminating an unintended pregnancy, carried out either by persons lacking the necessary skills or in an environment that does not conform to minimal medical standards, or both [1]. Every year, approximately 25 million unsafe abortions occur worldwide; of these, 97% are reported in developing countries, and half of them in Asia [2]. Unsafe abortion plays an important role in maternal morbidity, disability and mortality; largely from post-abortion sepsis, haemorrhage, genital trauma, infection and infertility [3]. Recent estimates suggest that about 13% of global maternal deaths are attributed to unsafe abortion [4]. Also, approximately seven million women undergo treatment due to complications from unsafe abortion [5]; and about five million women suffer disability as a result of such complications [6]. Because of high maternal morbidity, mortality, and disability caused by unsafe induced abortion, the 57th World Health Assembly endorsed unsafe abortion as a major public health concern in 2004; since then, eliminating unsafe abortion has become an important agenda for WHO global strategy on reproductive health [7]. This global strategy also suggested that eliminating unsafe abortion would require scientific evidence to formulate relevant policies and programs.

Globally, the underlying causes of unsafe abortion are unmet need for family planning and unintended pregnancy [8]. In developing countries, women often choose unsafe abortion services to end unintended pregnancies [3]. Unsafe abortion rate is estimated to be 16 per 1000 women in low- and middle- income countries, which is slightly lower than South-Central Asian region (estimated to be 17 per 1000 )[3]. Unsafe abortion rate and related complication are high when countries lack legal access to abortion and/or have no institutional provision for safe abortion services [9]. Studies conducted in LMICs of African and Latin American region reported that unsafe abortion rate was higher among women with lower income, ethnic minorities, and lower education [10-12].

In South Asia, Nepal has become a pioneer in legalization, implementation and scale-up of safe abortion services [13]. In 2002, the Nepalese government granted women the right to abortion up to a specific gestational age-dependent upon circumstances or medical conditions. For instance, women can terminate pregnancy on request within the first 12 weeks of gestation. In case of rape or incest, pregnancy can be terminated up to 18 weeks of gestation. If a doctor recommends that the pregnancy poses a danger to the life, physical or mental health of the pregnant woman, or if the fetus is seriously deformed, then abortion can be done any time of gestation [14]. Following this legal reform, a comprehensive safe abortion care program was implemented in 2004 [13]. In 2009, after the feasibility study of safe induced medical abortions services for pregnancies up to 9 weeks of gestational age, the phase-wise scaling up of the program was initiated in rural health posts with birthing centres facilities by skilled birth attendants (auxiliary nurse midwives having two months training on safe childbirth skills)[15, 16]. Until 2017, medical abortion services were available in 49 districts (out of 77 districts) [17, 18]. Abortion services are provided in certified health facilities by doctors and skilled birth attendants trained on abortion services [18-20].

After more than decade-long programmatic responses, the utilization of safe abortion services has not yet been universally adopted in Nepal. For instance, in 2014, out of total estimated 323,000 abortions, about 58% of abortions were conducted using a clandestine procedure provided by untrained/uncertified health providers or induced by the pregnant woman herself [19]. Previous literature has documented that there are challenges for the delivery of abortion services that include limited coverage of abortion sites, lack of trained human resources, and necessary equipment and medicines in accredited health facilities [19]. A lower contraceptive prevalence rate (53%) and higher unmet need for family planning services (24%) [21] resulted in high unintended pregnancy[21] that could potentially compel women to use unsafe abortion services. Also, a qualitative study in Nepal reported that abortion service seekers experienced denial from safe abortion services due to higher gestational age, and these women adopted unsafe abortion practices [22]. Women who sought abortion services had lower knowledge on the location of certified abortion sites[23] as well as legal conditions of abortion with higher unintended and untimed pregnancies [24-26].

Additionally, women who reported unsafe abortion were less likely to know the legal provision of abortion in Nepal compared with those who reported safe abortion services [25]. A recent study conducted in Nepal revealed that women of higher socioeconomic status had lower odds of unsafe abortion practices [27]. However, this study is insufficient to unpack the contributing factors for the needs of unsafe abortion practices, including knowledge on safe places for abortion services. There is a dearth of knowledge gaps in the role of enabling and modifiable factors that could be useful to revise the abortion policies and practices in Nepal.

This suggests the scientific evidence is needed to revisit the existing policies and programs for eliminating unsafe induced abortions practices in Nepal. The WHO suggests that empirical research on unsafe abortion would help to re-evaluate existing programs as well as formulate appropriate strategies to improve safe abortion services[1, 3, 28]. Hence, this study aimed to provide a national estimate on the unsafe abortion rate and examine factors associated with unsafe abortion using the data from the Nepal Demographic and Health Surveys (NDHS) 2011 [29] and 2016 [21]. The findings from this study would open up discussion around evaluating existing abortion policies and programs and designing targeted strategies to eliminate unsafe abortion and achieve the maternal health-related target of 3.2 of Sustainable Development Goals (SDGs) 3[30].

Methods

Data sources

This research has derived data from the NDHS 2011 and 2016 (available from https://dhsprogram.com/data/new-user-registration.cfm). NDHS is also part of the Demographic and Health Survey Program. The DHS program is US Government-funded global health program, provides technical and financial support to conduct demographic and health surveys and health facility surveys in more than 90 LMICs around the globe. These surveys are implemented in partnership with ICF International (USA based company) and the government of the host country. In Nepal, under the leadership of the Ministry of Health and Population and technical support from ICF international, New Era (local research organization) conducts the NDHS in every five years[21, 29, 31].

Sample

Data used in this analysis were based on women’s questionnaires. The NDHS used two-stage cluster random sampling. A total of thirteen strata were constructed using five development regions and three ecological regions. In the first stage, the primary sampling units, wards of rural and sub-wards of urban areas of each stratum were selected, which also called as Enumeration Areas (EAs). In the second stage, households were selected using simple random sampling technique. The details of sampling techniques are further described in the full report of NDHS 2011 and 2016.

The data of NDHS 2011 and 2016 were merged to get the maximum sample size for this study. A total of 25, 536 women of reproductive age (15–49 years) were interviewed in the two surveys (NDHS 2011 and 2016). The average response rates for women aged 15–49 years in the NDHS 2011 and 2016 were 98%. Women who received the most recent abortion services five years prior to the surveys constituted study population. A total of 911 women received abortion services during the survey period.

Outcome variable

In the surveys, information on the abortions services was collected using the following questions. In the pregnancy history section of the questionnaire, women were asked: Did you, or someone else do something to end this pregnancy?' has a yes/no response. Women responding ‘yes’ are then asked further questions about their abortion.

The outcome variable for this study was ‘unsafe abortion’. Based on WHO definition [1], unsafe abortion was coded as ‘1’ if the pregnancy was terminated either by persons lacking the necessary skills or in an environment that does not conform to minimal medical standards or both; otherwise coded as “0”. To comply with this definition, Nepal’s abortion law [13], and previous literature [27], this research considered unsafe abortions if conducted by other than physicians and nurse-midwives or those carried out outside health facilities.

Independent variables

Past studies conducted in Ghana [10], Ethiopia [12], Mexico [11] on factors associated with unsafe abortion, and the information available in datasets were employed as a basis for the selection of potential confounding variables [Fig 1]. Some variables such as ethnicity, wealth status, and knowledge of safe abortion place or legal conditions of abortion were further categorized for this study. For instance, the Government of Nepal has categorized ethnicities into six broad groups [32]: i) Dalit (Hill and Terai)); ii) Janajati (Indigenous Hill and Terai); iii) Madhesi (non-Dalit Terai caste groups); iv) religious minorities (Muslims); v) upper caste groups (Brahman/Chhetri) vi) Others (Thakuri and Sanayshi). Based on similar socioeconomic and geographical similarities, and other literature [33, 34] ethnicities were categorized into four groups: a) Brahman/Chhetri (merging with “Others” category); b) Dalit; c) Janajati; d) Madhesi (merging with “Muslims” category). Similarly, knowledge about certified abortion sites and legal conditions of abortion were categorised as: i) knew the legal condition of abortion and place for safe abortion, ii) only know the legal conditions, iii) only knows the location of the place for safe abortion, and iv) did not know both. In NDHS, wealth quintiles were calculated using principal component analysis of more than 40 households’ asset items. In this research, households’ wealth quintile were categorised into three groups: the bottom 40% was referred to as poor households, the next 40% as the middle households and the top 20% as rich households, consistent with previous studies [35, 36].

Fig 1

The conceptual framework of factors of unsafe abortion adapted from modified Anderson’s behavioural model [37].

Conceptual framework

A modified Anderson’s behavioural model of health service use [37], which has been consistently used in other studies [38, 39], was adopted as a conceptual framework for this analysis [Fig 1]. According to this model, predisposing, enabling and need factors contribute to use/non-use of any health services.

Fig 1 shows the predisposing, enabling, and need factors of unsafe abortion services. Predisposing factors are existing conditions (not directly responsible for use) that predispose women to use or not abortion services. In this study, place of residence, women’s age, the socioeconomic status including women education (women and their husbands), literacy status, ethnicity, gender (sex of the last-child), the total number of living son or daughters were considered as predisposing factors. Similarly, enabling factors are conditions that facilitate or impede the use of services. In this research enabling factors for unsafe abortion were household wealth index, occupation, mass media exposure, knowledge of legal conditions and certified abortion sites. Need factors are needs or conditions that women compel to use the services. In this study, the unmet need for family planning or unintended pregnancy, women’s reasons for abortions, and gestational age at abortion were considered as need factors [Fig 1].

Statistical analysis

Statistical analysis employed descriptive and staged regression models. Firstly, descriptive statistics such as frequencies and proportions were calculated to provide population-based estimates of the outcome variable. Abortion rates were calculated considering the definition of total number of abortion (safe or unsafe abortion) occurring in a specified period per 1,000 women aged15-49 years [3]. This research estimated the rates of abortion and unsafe abortion and their 95% Confidence Interval (CI). Secondly, staged logistic regression [40-42] models were conducted to examine factors associated with unsafe abortion while adjusting for potential confounding factors. Unadjusted odds ratios and their 95% CI were calculated to examine the association between each independent variable and unsafe abortion (model 1).

Before moving to the multivariate logistic regression analysis, multi-collinearity was checked using variation inflation factors (VIF) test considering VIF cut-off value >3[43] (none of the independent variables was found cut-off values> 3). At the second stage, the predisposing factor was entered and used manual backward elimination technique to retain statistically significant variables associated with unsafe abortion at 5% significance level (model 2). The same procedure was followed when enabling, and need factors added in the third stage (model 3), and the final stage (model 4), respectively. Factors significantly associated (p<0.05) with unsafe abortion in the final model (model 4) was reported [34]. To confirm/validate the result of the staged regression model, other alternative logistic regressions were also conducted [34, 36] a) entering only potential risk factors with p-value < 0.20 obtained in the bivariate analysis for backward elimination process, and b) testing the backward elimination method by including all potential risk factors. Complex sample analyses technique was adopted throughout to account for the study design, and sample weight, and analysis [36, 44]. A total of 45 missing values were excluded from the multivariate analysis. All analyses were performed in STATA (Stata Corp, College Station, Texas US) software version 14.0.

Ethics approval

These surveys were approved by an ethical review board of Nepal Health Research Council, Nepal, and ICF Marco International, Maryland, USA. The first author got permission from DHS program (USA) to use those datasets for this study.

Results

Descriptive characteristics of the study population

Out of 911 women who used abortion services during 2011–2016, slightly over 50% were living in rural areas [Table 1]. Overall, 50% of the women and 72% of their husbands had secondary and higher level of education. Having access to general mass media and knowledge of safe abortion place were almost equally distributed (91% and 90% respectively).

Table 1

Descriptive characteristics of the study population and the proportion of unsafe abortion in Nepal, 2011–2016 (N = 911).

Variables	Categories	Total abortion	Unsafe abortion (%)	P
Total population		911	236 (26)
Predisposing factors
Rurality	Rural	495	107(22)	0.008
	Urban	416	129(31)
Eco-region	Hill	419	93(22)	0.035
	Terai	438	122(28)
	Mountain	54	21(39)
Development region	Western	268	62(23)	0.193
	Central	240	58(24)
	Eastern	153	38(25)
	Mid-western	136	48(36)
	Far-western	115	30(26)
Women’s age	34–49 years	216	51(24)	0.664
	20–34 years	652	174(27)
	<20 years	43	11(24)
Ethnicity	Brahmin/Chettri	402	84(21)	<0.001
	Dalit	119	45(38)
	Janajati	281	65(23)
	Madhesi and Muslim	109	42(38)
Women’s education level	Secondary or higher	459	97(21)	0.012
	Primary	216	65(30)
	No education	236	74(31)
Women’s literacy level	Can read part or whole of the sentence	719	169(23)	0.004
	Cannot read	192	67(35)
Numbers of male children	None	216	41(19)	0.054
	One	417	112(27)
	Two or more	278	84(30)
Numbers of female children	None	290	87(30)	0.035
	One	337	69(21)
	Two or more	284	79(28)
Sex of the most recent child	Male	509	149(29)	0.043
	Female	361	80(22)
Husband education	Secondary or higher	659	148(23)	<0.001
	Primary	168	59(35)
	No education	76	28(36)
Enabling factors
Wealth index	Rich	206	33(16)	0.002
	Middle	376	95(25)
	Poor	329	108(33)
Women’s occupation	Skilled	262	46(18)	0.007
	Agriculture	418	124(30)
	Not working	231	66(28)
Women’s working status	Currently working	579	146(25)	0.591
	Currently not working	332	90(27)
Exposure to general mass media	No	80	36(44)	<0.001
	Yes	831	200(24)
Exposure to mass media on public health issues	No	174	66(38)	<0.001
	Yes	737	170(23)
Need factors
Unmet need for family planning	No unmet need	602	155(26)	0.861
	Unmet need	309	81(27)
Knowledge of condition and place of safe abortion	Knows condition and place for safe abortion	610	131(21)	<0.001
	Knows condition only	57	36(63)
	Knows place only	212	49(23)
	Absence of both	32	20(62)
Reason for abortion	Health of women	94	14(15)	<0.001
	Wanted to delay/spacing	174	57(32)
	Unwanted birth	403	127(32)
	Low family earning and others£	240	38(16)
Gestation(N = 735)	Up to 8 weeks	580	150(26)	0.583
	9–12 weeks	117	26(22)
	13 weeks and more	38	8(21)

P-value obtained from Chi-square association

Abortion practices

Out of 25,536 women surveyed during the period (2011–2016), 911 women used abortion services; and of these abortion services, 23% (236) were unsafe abortions. The rate of abortion was estimated as 36 (95% CI: 33, 38) per 1000 women aged 15–49; whereas the rate of unsafe abortion was seven (95% CI: 6, 8) per 1000 women aged 15–49 years [Table is not shown].

Descriptive analysis of unsafe abortion

The majority (17%) of the abortions were below eight weeks of gestational age(Table 1). A substantial proportion of unsafe abortions were conducted in the Mountain region (39%), and among those with the disadvantaged ethnic background (Dalit, and Madheshi and Muslim). Similarly, a higher proportion of women were found to undertake unsafe abortion practices if they or their husbands reported no education (36%), if they could not read or write, belonged to the households of lower wealth index, or were involved in agricultural occupation. If women had lower knowledge of legal conditions and safe abortion places (62%), or if they had no exposure to mass media, then a higher proportion of women used unsafe abortion services. If women wanted to delay or space childbirth or did not want birth, then a higher proportion of women were found to use unsafe abortion services [Table 1].

Factors associated with unsafe abortion practices in Nepal

Table 2 shows the results of bivariate and multivariate regression analyses of independent variables and unsafe abortion in Nepal. The bivariate logistic regression showed that rurality (urban), eco-region (Mountain), development region (mid-western), wealth index (middle or poor), ethnicity (Dalit, or Madhesi and Muslim), maternal education (primary or no education), women’s literacy level (cannot read), husband education (primary or no education), maternal occupation (agriculture or no occupation), knowledge on legal conditions of abortion and safe abortion sites, exposure to general mass media (yes), and exposure to mass media on public health issues, number of male children (≥ 2), number of female children (one), sex of the most recent children (female), reasons for abortion (want to delay/space child-bearing, or unwanted child) were all significantly associated with unsafe abortion at p<0.05 [Table 2].

Table 2

Unadjusted and adjusted odds ratio of factors associated with unsafe abortion in Nepal in 2011–2016 (N = 911).

Variables	Categories	Unadjusted OR (95% CI)	P	Adjusted OR (95% CI)	P
Predisposing factors
Rurality	Rural	1.00		1.00
	Urban	1.63(1.13, 2.36)	0.009	2.11 (1.37, 3.24)	<0.03
Eco-region	Hill	1.00		1.00
	Terai	1.35(0.91, 2.00)	0.140	1.47(0.98, 2.21)	0.063
	Mountain	2.22(1.27, 3.88)	0.005	2.36(1.21, 4.60)	0.012
Development region	Western	1.00
	Central	1.04(0.60, 1.81)	0.890
	Eastern	1.10(0.61, 1.97)	0.747
	Mid-western	1.84(1.11, 3.02)	0.017
	Far-western	1.51(0.64, 2.07)	0.637
Predisposing factors
Women’s age	34–49 years	1.00
	20–34 years	1.18(0.80, 1.75)	0.399
	<20 years	1.04(0.48, 2.27)	0.914
Ethnicity	Brahmin/Chettri	1.00		1.00
	Dalit	2.32(1.32, 4.07)	0.004	1.89 (1.02, 3.52)	0.043
	Janajati	1.13(0.76, 1.70)	0.535	1.35 (0.90, 2.03)	0.146
	Madhesi and Muslim	2.37(1.45, 3.86)	0.001	2.10 (1.25, 3.54)	0.005
Women’s education level	Secondary or higher	1.00
	Primary	1.60(1.05, 2.43)	0.028
	No education	1.71(1.15, 2.57)	0.009
Women’s literacy level	Can read part or whole of the sentence	1.00
	Cannot read	1.74(1.19, 2.54)	0.004
Husband education	Secondary or higher	1.00		1.00
	Primary	1.87(1.20, 2.91)	0.006	1.72(1.07, 2.75)	0.024
	No education	1.98(1.12, 3.48)	0.018	2.12(1.06, 4.22)	0.033
Numbers of male children	None	1.00
	One	1.58(0.95, 2.61)	0.076
	Two or more	1.75(1.08, 2.83)	0.023
Numbers of female children	None	1.00
	One	0.63(0.41, 0.98)	0.040
	Two or more	0.88(0.58, 1.33)	0.536
Sex of the most recent child	Male	1.00
	Female	0.70(0.49, 0.99)	0.042
Enabling factors
Wealth index	Rich	1.00
	Middle	1.75(1.00, 3.03)	0.047	1.70(0.91, 2.87)	0.112
	Poor	2.52(1.50, 4.24)	0.001	2.16 (1.18, 3.94)	0.043
Women’s occupation	Skilled	1.00		1.00
	Agriculture	1.94(1.25, 3.01)	0.003	1.82(1.16, 2.86)	0.009
	Non- agriculture	1.84(1.18, 2.88)	0.008	1.53(0.93, 2.50)	0.092
Women’s working status	Currently working	1.00
	Currently not working	1.09(0.79, 1.52)	0.592
Exposure to general mass media	No	1.00
	Yes	0.40(0.24, 0.66)	<0.001
Exposure to mass media on public health issues	No	1.00
	Yes	0.49(0.33, 0.71)	<0.001
Need factors
Unmet need for family planning	No unmet need	1.00
	Unmet need	1.03(0.71, 1.50)	0.862
Knowledge of condition and place of safe abortion	Knows condition and place for safe abortion	1.00		1.00
	Knows legal conditions but not place	6.34(3.41, 11.77)	<0.001	5.13(2.64, 9.98)	<0.001
	Knows place but not legal conditions	1.10(0.73, 1.65)	0.652	1.34(0.88, 2.03)	0.172
	Absence of both	6.00 (2.81, 12.81)	<0.001	4.83(2.20, 10.61)	<0.001
Reason for abortion	Health of women	1.00		1.00
	Wanted to delay/spacing	2.75(1.43, 5.32)	0.003	2.71(1.39, 5.28)	0.003
	Unwanted birth	2.66(1.36, 5.19)	0.004	2.33(1.19, 4.56)	0.014
	Low family earning and others£	1.08(0.53, 2.19)	0.831	1.36(0.64, 2.89)	0.418

Bold values indicate significance in the final model at p<0.05.

£ Others category also include a reason such as no one in the family to look after the child, and to avoid shame.

The final regression model [Table 2] revealed that women residing in the mountain region (aOR 2.36 95% CI 1.21, 4.60), or rural women (aOR 2.11, 95% CI 1.37, 3.24) were predisposed to unsafe abortion compared their hill or urban peers [Table 2].

Enabling factors such as women belonging to poor household had higher odds of having unsafe abortion (aOR 2.16, 95% CI 1.18, 3.94) compared to women of wealthy households. Additionally, unsafe abortion were significantly higher among Dalit (aOR 1.96, 95% CI 1.08, 3.54), Madhesi or Muslims (aOR 1.71, 95% CI 1.01, 2.88) compared to Brahmin/ Chhetri ethnic group. Husbands with no education (aOR 2.12 95% CI 1.06, 4.22), and women having occupation in agricultural sector (aOR 1.82 95% CI 1.16, 2.86) had higher odds of unsafe abortion compared to husband with secondary and higher education and women with skilled occupation respectively [Table 2].

Need factors such as knowledge on safe abortion places and legal conditions, and reasons for abortions were also significantly associated with unsafe abortion practices in Nepal. Women who did not know the place for safe abortion services (aOR 5.13 95% CI 2.64, 9.98) (but know legal conditions of abortions), and who did not know both (legal conditions of abortions and place for safe abortion) had higher odds of unsafe abortion practices compared with those who did know both. Finally, women who had unwanted pregnancy or wanted to delay or space childbirth had higher odds of unsafe abortion practices [Table 2].

Discussion

This study revealed that the rates of abortion and unsafe abortion over the study period (2011–2016) were 36 and seven per 1000 women aged 15–49 years respectively. Independent variables such as eco-region, rurality, ethnicity, wealth index, husband education or women’s occupation, knowledge on legal conditions of abortions and place for safe abortion, reasons of abortion were significantly associated with unsafe abortion.

The higher risk of unsafe abortion in the Mountain region may be aggravated due to difficult geographic terrain that may hamper both the access and utilization of safe abortion services. Availability of abortion services is limited to district hospitals or primary health care centres in the mountainous districts. Though medical abortion services have been available up to the health post level (birthing centre- health post having childbirth facilities only), many mountainous districts have not been covered by medical abortion services [32]. Women have to spend several hours to reach health facilities to get safe abortion services [20]. In addition, even health facilities are certified as abortion sites, unavailability of trained human resources, equipment, drugs are other challenges that bar safe abortion services in the Mountainous region could be the challenge [20]. In agreement with previous studies conducted in Nepal [27] and Tanzania [44], this study found that women living in rural Nepal were at higher risk of unsafe abortion.

Compared to other ethnic and religious groups, abortion practices are religiously stigmatized in Muslim communities, and culturally taboo in Madhesi and Dalits[19, 45]; and post-abortion women are often labelled as sinners (Papini), ill-luck (alichhini), murderers (jyanmaara), and foetus killers (garbhaghati) [19]. The higher odds of unsafe abortion amongst Muslim women in this study may be due to these cultural barriers that make women use abortion services other than certified health facilities or trained providers. In Nepal, the contraceptive prevalence rate is low; whereas, the unmet need for family planning is high [46]. The lower contraceptive prevalence rate and the higher unmet need for family planning are considered as important contributors to unwanted pregnancy-a possible reason for unsafe abortion as documented in public health literature [8]. In Nepal, people from Dalit ethnic background and those who live in the Terai are relatively poor that makes access to safe abortion services further hard as the provision of free abortion services is not yet universal in Nepal [20].

This study identified significant differences in unsafe abortion practices based on different socioeconomic status. For instance, women having occupation in agricultural sectors, husbands with no education background, and women belonging to the households of lower wealth quintile were all significantly associated with unsafe abortion. These findings were similar to the studies conducted in Brazil [47] and Mexico [11], which also found that unsafe abortion was higher among women of lower-income, and women with low-level education. In Mexico, the legal status of abortion varied by state; Mexico city offers abortion up to 13 weeks gestation, whereas in Brazil abortion is legal if pregnancies result from rape or incest or if the life of the pregnant woman or fetus is at risk [48]. Both studies argued that the legal barriers to safe abortion services meant poor women could not afford quality abortion services, and they were compelled to use unsafe induced abortion. However, in Nepalese context, higher unsafe abortion practice among women of lower wealth status might be the financial inaccessibility to the safe abortion services as it was only made free of cost after 2017 [20]. Women from poor households were not able to get safe abortion services as women were required to pay at least 800–1200 Nepalese Rupees (8–12 USD) as service charge excluding medications (until data collection for NDHS 2016) [20]. In addition to the direct cost of abortion services, women are also required to pay other indirect costs such as cost for medicine, transportation, meal and accommodation [19, 20].

In contrast to conditions of Brazil [47] and Mexico [11], Nepal has overcome the legal barrier, but higher unsafe abortion is more prevalent among poor women. Higher unsafe abortion among poor socioeconomic groups in this study may be due to the need for family planning services. Socioeconomically disadvantaged and ethnic minorities groups in Nepal have lower contraceptive prevalence rates and higher unmet need for family planning services [21, 49]. Poor access and utilization of family planning services may lead to the use of abortion services as methods of spacing or delaying childbirth. However, women may not know the authorized place and legal conditions for abortion services [26], which possibly lead to unsafe abortion services.

The current study identified that women who did not know the place of safe abortion, regardless of their knowledge on legal conditions to have an abortion, had a higher likelihood of unsafe abortion practices. Previous studies conducted in Nepal revealed that women who were not aware of the legal provision (such as aborting period) or location of nearest safe abortion sites [23, 26] were more likely to have unsafe induced abortion. These facts show that being aware of certified abortion sites is important for the uptake of safe abortion services in Nepal.

In this study, though unmet need for family planning services was not significantly associated with unsafe abortion, the higher odds of unsafe abortion practices were significantly associated with child spacing or unwanted pregnancy. This indicates the need for family planning services to prevent unintended pregnancy. In Nepal, 24% of women had an unmet need for family planning (16% want to delay, and 8% want to space the birth), and 19% childbirth is from unwanted [21]. Evidence from Ghana suggests that unsafe abortion were higher if women have an unintended pregnancy [8]. Therefore, strengthening family planning services and reducing unintended pregnancy could be one of the strategies for reducing unsafe abortion in Nepal.

This study has some strengths and limitations. We pooled the data from nationally representative surveys conducted in the past decade. Thus, estimates from this study are generalizable to the Nepalese population and can inform national policies and practices. Secondly, the response to the surveys was high (>98%), reducing a likely chance of selection bias from the observed findings. However, there might be recall bias because the information was collected through the recall of past experiences, and the recall period was long (5 years) that many increase the potential for misclassification of cases. Due to the small sample size, this study could not do a separate analysis for each of the survey wave (NDHS 2011 and NDHS 2016) for absolute comparison. It is an analysis of quantitative data and lacks qualitative information to explain the behaviour of women. Hence, future qualitative studies are needed to explore more inclusive intervention for culturally diverse population across the country.

Policy and program implications

This study has policy and program implications. The legalization of abortion was the first move, but that does not seem sufficient enough for the delivery and utilization of safe abortion services[50]. Therefore, the increase in certified safe abortion sites and the provision of safe abortion services for women of the Mountainous region and socioeconomically disadvantaged groups could be an appropriate step to reduce unsafe abortion practices. From the demand side perspective, the community needs to be informed and sensitised about the use of safe abortion services[51]. Moreover, the integration of awareness-raising interventions in existing health programs could increase the demand for safe abortion services[52].

Unsafe abortion was higher in women with the lowest wealth status or women having occupation in the agricultural sector. For those groups, financial barriers could be a factor in the choice of unsafe abortion practices. The Government of Nepal has already made all abortion services freely available since 2017[20], but this might not be enough as users must pay for the cost of medicines. Just making services free may not address all the financial barriers for socioeconomically disadvantaged women, and abortion-related direct and indirect costs also need to be addressed while implementing abortion services. Given the findings that women using unsafe abortion practices to end unwanted pregnancy or space or delay childbearing, strengthening family planning service to the wider community is another vital strategy that may help to reduce unsafe abortion practices in Nepal.

Conclusion

Several factors contribution to unsafe abortion in Nepal. Availability of safe abortion services by establishing safe abortion sites could reduce unsafe abortion practices. Reduction of unintended pregnancy by use of family planning commodities may help women not to choose unsafe abortion practices as a method of child space or delay childbearing. Programmatic efforts should be focussed on access to abortion services to the Mountainous Region, among poor, Dalit and Madhesi and Muslim communities.

18 Jul 2019

PONE-D-19-14863

Factors associated with unsafe induced abortion practices in Nepal: Pooled Analysis of the 2011 and 2016 Nepal Demographic and Health Surveys.

PLOS ONE

Dear Dr. Khatri,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

We would appreciate receiving your revised manuscript by 18 August 2019. When you are ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter.

To enhance the reproducibility of your results, we recommend that if applicable you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

Please include the following items when submitting your revised manuscript:

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). This letter should be uploaded as separate file and labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. This file should be uploaded as separate file and labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. This file should be uploaded as separate file and labeled 'Manuscript'.

Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.

We look forward to receiving your revised manuscript.

Kind regards,

Russell Kabir, PhD

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

http://www.journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and http://www.journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We suggest you thoroughly copyedit your manuscript for language usage, spelling, and grammar. If you do not know anyone who can help you do this, you may wish to consider employing a professional scientific editing service.

Whilst you may use any professional scientific editing service of your choice, PLOS has partnered with both American Journal Experts (AJE) and Editage to provide discounted services to PLOS authors. Both organizations have experience helping authors meet PLOS guidelines and can provide language editing, translation, manuscript formatting, and figure formatting to ensure your manuscript meets our submission guidelines. To take advantage of our partnership with AJE, visit the AJE website (http://learn.aje.com/plos/) for a 15% discount off AJE services. To take advantage of our partnership with Editage, visit the Editage website (www.editage.com) and enter referral code PLOSEDIT for a 15% discount off Editage services.  If the PLOS editorial team finds any language issues in text that either AJE or Editage has edited, the service provider will re-edit the text for free.

Upon resubmission, please provide the following:

The name of the colleague or the details of the professional service that edited your manuscript

A copy of your manuscript showing your changes by either highlighting them or using track changes (uploaded as a *supporting information* file)

A clean copy of the edited manuscript (uploaded as the new *manuscript* file)

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Yes

Reviewer #4: Partly

Reviewer #5: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: Yes

Reviewer #5: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: No

Reviewer #3: Yes

Reviewer #4: No

Reviewer #5: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Good Effort

Please avoid the word "we or our". You can replace with "This research".

Please add reference number of THREE approval you obtained from Nepal Health Research Council, Nepal; and ICF Marco International Maryland, USA; and DHS program (USA).

Regards

Reviewer #2: Introduction

The paper aimed to (1) provide a national estimate on unsafe abortion rate in Nepal and to (2) examine the factors associated with unsafe induced abortion. Data derived from 2011 and 2016 Nepal Demographic and Health Surveys were analyzed using logistic regression. The authors reported that “women living in mountain Region, urban, poor households; disadvantaged ethnicities such as Dalit or non-Dalit Terai caste, and Muslim, involved in the agricultural sector had higher odds of having unsafe induced abortions compared to their reference’s groups. Women who did not know the location of safe abortion sites had higher odds of unsafe induced abortion, regardless of their knowledge of the legal conditions of abortion,” that “women who lacked knowledge of both place and legal conditions had higher odds of unsafe induced abortions compared to those who knew both,” and that “women who wanted to delay or space childbirth or unwanted birth were associated with higher odds of unsafe induced abortion.”

Merits

While this reviewer has specific comments with regards to some sections, the research on which this article is based is of importance for developing new abortion policies, and once the article is revised, some interesting findings may be gleaned from the data.

Remarks

However, there are a number of general and specific issues that require attention.

Firstly, although the manuscript is generally clear, it would benefit from rephrasing and remolding for clarity and style.

Secondly, the results in the abstract must be reported following specific guidelines established by the international scientific community. For instance, there is no mention of OR, 95% CI, and p values in the abstract.

Further, the authors have placed emphasis on the following factors: place of residence/region, women's age, education (women and their husbands), literacy status, ethnicity, gender (sex of the last child), total number of living male or female children, wealth status, occupation, exposure to mass media on public health issues, knowledge of legal conditions, knowledge of safe abortion place, unmet need for family planning, unintended pregnancy, women's reasons for abortions, and gestational age at abortion.

That is too many variables, some of which may be unrelated to the dependent variable. Though the manuscript attempts to address each of the above-mentioned factors, it failed to take into account the downsides of having models with many independent variables to select from. It is well-known that each irrelevant variable included in the model(s) will decrease the precision of the estimated parameters.

Given the high number of potential predictor variables, it would have been better if the authors had selected the forward stepwise regression (instead of the backward elimination technique used in this paper). The forward stepwise regression is recommended when having a large set of potentially relevant predictor variables. It generates a good sequence of models by allowing to fine-tune them to obtain important information about the quality of the potential predictors.

The backward elimination technique used by the authors is usually applied when there is a modest number of potential predictors, which was not the case here.

Furthermore, the limitation section lacks to mention the limitations of applying the backward elimination method in the selection of potentially relevant variables included in the regression models. Additionally, it is unclear whether the authors used cross-validation to detect potential cases of overfitting and collinearity.

Apart from that, some of the results need to be presented in a different manner, and it is recommend to add more figures/graphs.

Finally, the authors mention in the discussion section that “Among poor, and disadvantaged ethnic communities in Nepal, the contraceptive prevalence rate is high.” They then go on and state that “socioeconomically and disadvantaged ethnic groups have lower contraceptive prevalence rates.” The paper needs coherence.

I hope this review will be helpful and wish the authors the best of luck with their research!

Reviewer #3: This manuscript addresses a relevant topic, such as the determinants of unsafe abortions. It is easy to read and well written. My only concern is about the pooled analysis. Although it is probably necessary in order to obtain a sample big enough, it seems that the main number of unsafe abortions belong to year 2007. As one of the objectives of the study is to propose policies in order to reduce unsafe abortions, conclusions obtained could correspond to the profile of unsafe abortions in 2007, more than in the present. So, I suggest to repeat the analyses conducted in table 2 also in a separate way for each of the years analysed, in order to explore if there are any differences for this period.

Reviewer #4: Congratulations on your work to generate evidence on unsafe abortion practices in Nepal. This cross-sectional study aimed to examine the factors associated with unsafe induced abortion practices in Nepal using 2011 and 2016 Nepal Demographic and Health Surveys. The findings of the study may be useful for policy makers, however, I have some concerns regarding the statistical analysis and discussion of results. In addition, the manuscript needs to be reviewed by a professional English native editor. Some sentences are incomplete or unclear. Please find the detailed comments below by each section.

Introduction:

Overall, introduction needs to be revised. The authors tried to provide data on unsafe abortion at global and national level, however, the authors could present more in greater depth regarding what current evidence is (what do we know now), what is the gap and how this study will fill this gap. The authors need to conduct a proper literature review to provide up to date studies on this topic. The authors could indicate global perspective and findings of other previous studies investigating factors associated with unsafe induced abortions. Later, the authors could mention relevant studies conducted in Nepal and the gaps needed to be addressed. The authors said that there is no study conducted at the national level, however, the authors could mention relevant studies conducted at communication level in Nepal to provide a summary of findings from previous studies.

The authors mentioned that ‘Some studies reported that unsafe abortion rate was higher among women with lower income, ethnic minorities, and lower education’. This looks similar as the finding of this study. Please clearly mention what are the added value of this study.

There is no justification why the authors used 2011 and 2016 NDHS. Please mention why the authors did not use 2001 DHS or 2014 MICS (Multiple Indicator Cluster Surveys).

What is the reference of the sentence ‘The WHO suggests that empirical research on unsafe abortion would help to re-evaluate existing programs as well as formulate appropriate strategies to improve safe abortion services.’? In addition, this sentence does not strengthen the justification of this survey because it is not an empirical research.

Methods:

The authors should provide more details in statistical analysis.

Data source and sample:

1. I suggest the authors to describe DHS in general.

2. what is the total sample size in the end? How did you come to this final sample size?

3. How did you handle missing data?

Independent variables:

For ethnicity, the authors merged some ethnic groups with small sample size into other ethnic groups and said that these groups were similar each other. However, the authors did not provide any evidence with reference on this. Moreover, the authors need to indicate the number of sample size of certain ethnic groups instead of saying ‘small size’. How about Newari origins? This is also one of the unique and major ethnic groups in Nepal.

‘husband education’ is mentioned twice in the Fig 1. conceptual framework. Please remove one.

Statistical analysis:

1. The authors conducted a four staged multivariate logistic regression model but they did not explain why this method is the best to achieve the goal of the study

2. The authors mentioned unadjusted odds ratios as (aOR). Did you mean adjusted odds ratios?

3. The authors did not mention how to choose reference groups when performing logistic regression analysis. Please explain.

4. There are too many independent variables and some are highly interrelated such as women’s education and literacy and women’s occupation and working status. Have the authors checked multicollinearity?

Results:

Descriptive characteristics of the study population:

1. Table 1:

a. It is not clear to me what chai-square means here- is it for categories under unsafe abortion? Please specify.

b. I suggest the authors to revise the table 1. Column percentage and row percentage are mixed so it is confusing.

c. The authors used the symbol, “@ and *”. Need to check whether this is in line with the PLOS ONE guideline.

Unsafe abortion practices in Nepal:

1. The authors mentioned that ‘Over the study period (2007-2016) in Nepal, the total and unsafe abortion rates were 36 (95% CI: 33, 38) and seven (95% CI: 6, 8) per 1000 women aged 15-49 years respectively’. However, there is no table or figure with this data. The authors should present the results with tables or figures. If not in the main manuscript, the authors could provide data in supplementing document.

2. Saying ‘study period 2007-2016’ is confusing. Suggest revising as ‘Data from 2011-2016 NDHS’.

3. The authors mentioned methods of unsafe induced abortions (medical, surgical, etc.), however, there is no data in the table. The authors should present all mentioned data in the table or figure.

Factors associated with unsafe abortion in Nepal:

1. Table 2

a. Why there are three empty rows under predisposing factors?

b. It is not easy to understand the table 2. The authors can consider presenting the results with figure to have a better visualization of results.

2. Even though the authors indicated that they conducted a four staged multivariate logistic regression model, there is no results of model 1-4. What are the results?

3. Also, there is no results regarding this sentence ‘To avoid any statistical bias, the results from the staged model were also checked by: (1) entering only potential risk factors with p-value < 0.20 obtained in the univariate analysis for backward elimination process, and (2) testing the backward elimination method by including all potential risk factors’. It is not clearly mentioned.

Discussion:

In general, the authors should provide a greater explanation of the findings in the discussion section. For instance, it is not clear the implication of the sentence ‘In Mexico, the legal status of abortion varied by state; Mexico City offers abortion up to 13 weeks gestation, whereas in Brazil abortion is legal if pregnancies result from rape or incest or if the life of the pregnant woman or fetus is at risk’. Why is it important and what needs to be done to improve the situation?

Implication:

It would be good to provide relevant references regarding the arguments of the authors. For instance, is there any studies supporting the sentence ‘From the demand side perspective, the community needs to be informed and sensitised about the use of safe abortion services. Moreover, the integration of awareness raising interventions in existing health programs could increase the demand for safe abortion services.’? It may help make the argument strong.

Reviewer #5: Abstract: Can be made more concise

Methods: How was the wealth index calculated? What was the assessment tool used by the NDHS?

How was the rates calculated for abortion rates and unsafe abortion rates?

How was it ensured that the health practitioners who did the abortion were certified for it?

Results: Redesign the table 1 and 2. Make it more clear.

Check the numbers, there are discrepancies. If it is a case of missing data, justify

Limitation and biases has to me mentioned

Discussion needs some more papers which could be more contextual in the countries setting.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Dr Mainul Haque

Reviewer #2: No

Reviewer #3: Yes: Isabel Aguilar

Reviewer #4: No

Reviewer #5: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files to be viewed.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at figures@plos.org. Please note that Supporting Information files do not need this step.

6 Aug 2019

Point by point responses to the reviewers’ comments

Reviewer #1: Good Effort

Thank you so much for appreciating our work.

Comment: Please avoid the word "we or our". You can replace with "This research".

Please add reference number of THREE approval you obtained from Nepal Health Research Council, Nepal; and ICF Marco International Maryland, USA; and DHS program (USA).

Response: Agreed, and changed in the revised manuscript as suggested. Regarding approval, we used publicly available secondary data obtained from the DHS program (https://dhsprogram.com/data/available-datasets.cfm). The first author sought approval from MEASURE DHS by online application form to use the data for this study. The details for the application process can be found in the link below: https://dhsprogram.com/data/using-datasets-for-analysis.cfm

Reviewer #2: While this reviewer has specific comments with regards to some sections, the research on which this article is based is of importance for developing new abortion policies, and once the article is revised, some interesting findings may be gleaned from the data.

Thank you for praising our manuscript; and we are pleased to address each of the reviewer’s comments as listed below.

However, there are a number of general and specific issues that require attention.

Thank you for pinpointing important general and specific issues which we have tried to address our best to satisfy the reviewer’s concern.

Comment: Firstly, although the manuscript is generally clear, it would benefit from rephrasing and remolding for clarity and style.

Response: Rephrasing and remodeling have been offered as required.

Comment: Secondly, the results in the abstract must be reported following specific guidelines established by the international scientific community. For instance, there is no mention of OR, 95% CI, and p values in the abstract.

Response: Thanks. Corrected as suggested (Please see the results section of abstract of the revised manuscript).

Comment: Further, the authors have placed emphasis on the following factors: place of residence/region, women’s age, education (women and their husbands), literacy status, ethnicity, gender (sex of the last child), total number of living male or female children, wealth status, occupation, exposure to mass media on public health issues, knowledge of legal conditions, knowledge of safe abortion place, unmet need for family planning, unintended pregnancy, women’s reasons for abortions, and gestational age at abortion.

That is too many variables, some of which may be unrelated to the dependent variable. Though the manuscript attempts to address each of the above-mentioned factors, it failed to take into account the downsides of having models with many independent variables to select from. It is well-known that each irrelevant variable included in the model(s) will decrease the precision of the estimated parameters.

Given the high number of potential predictor variables, it would have been better if the authors had selected the forward stepwise regression (instead of the backward elimination technique used in this paper). The forward stepwise regression is recommended when having a large set of potentially relevant predictor variables. It generates a good sequence of models by allowing to fine-tune them to obtain important information about the quality of the potential predictors.

Response: Thank you for the comment. The variables included in this study are important socio-demographic, and maternal factors that are widely used in public health literature as potential predictor variables [1-3]. Given due importance, information on these variables are found in NDHS maternal data file; hence, why included to examine any possible association with the outcome variable.

The backward elimination technique used by the authors is usually applied when there is a modest number of potential predictors, which was not the case here.

Furthermore, the limitation section lacks to mention the limitations of applying the backward elimination method in the selection of potentially relevant variables included in the regression models. Additionally, it is unclear whether the authors used cross-validation to detect potential cases of overfitting and collinearity.

Response: We consulted this with a mathematical and applied statistician. We kindly disagree with the reviewer, and we think the reviewer meant the contrary because ‘Overfitting’ occurs when a model is having too many parameters (variables). The staged model was introduced in this study to avoid the issue of overfitting. Also, we also tested our stage modelling approach by using both forward and backward elimination method; the three methods found the same variables to be significantly associated with unsafe abortion. We have also tested and reported multi-collinearity (please see last paragraph of page 8, statistical analysis section. Our approach of statistical analysis is consistent with previous studies [4-6].

Apart from that, some of the results need to be presented in a different manner, and it is recommend to add more figures/graphs.

Response: Our apology. We would be grateful if the reviewer can be more specific on his/her comment that which results are recommended to be in the figure. However, the way we have presented our results are easy for readers to navigate. In addition, this style of presenting the results are widely found in recent public health literature.

Finally, the authors mention in the discussion section that “Among poor, and disadvantaged ethnic communities in Nepal, the contraceptive prevalence rate is high.” They then go on and state that “socioeconomically and disadvantaged ethnic groups have lower contraceptive prevalence rates.” The paper needs coherence.

Response: Thank you for this mistake, we have corrected this.

I hope this review will be helpful and wish the authors the best of luck with their research!

Response: We are grateful to the reviewer; and we have addressed almost all the comments from reviewer 2 while taking other reviewers comments into account. Thank you very much for wishes.

Reviewer #3: This manuscript addresses a relevant topic, such as the determinants of unsafe abortions. It is easy to read and well written. My only concern is about the pooled analysis. Although it is probably necessary in order to obtain a sample big enough, it seems that the main number of unsafe abortions belong to the year 2007. As one of the objectives of the study is to propose policies in order to reduce unsafe abortions, conclusions obtained could correspond to the profile of unsafe abortions in 2007, more than in the present. So, I suggest repeating the analyses conducted in table 2 also in a separate way for each of the years analysed, in order to explore if there are any differences for this period.

Response: We agree with the reviewer, and the aim of using pooled datasets was to increase the sample size so as to increase the statistical power to help detecting any statistical differences in the course of statistical modelling; consistent with previous studies [4-6]. As per the reviewer’s suggestion, we have however accommodated this as a limitation of the study (Please see in the revised manuscript which reads as: ‘Due to the small sample size, this study could not do a separate analysis for each of the survey wave (NDHS 2016 and NDHS 2011) for absolute comparison’).

Reviewer #4: Comment: Congratulations on your work to generate evidence on unsafe abortion practices in Nepal. This cross-sectional study aimed to examine the factors associated with unsafe induced abortion practices in Nepal using 2011 and 2016 Nepal Demographic and Health Surveys. The findings of the study may be useful for policy makers, however, I have some concerns regarding the statistical analysis and discussion of results. In addition, the manuscript needs to be reviewed by a professional English native editor. Some sentences are incomplete or unclear. Please find the detailed comments below by each section.

Response: the manuscript has been reviewed by a professional editor. Language has been edited as suggested.

Introduction:

Comment: Overall, introduction needs to be revised. The authors tried to provide data on unsafe abortion at global and national level, however, the authors could present more in greater depth regarding what current evidence is (what do we know now), what is the gap and how this study will fill this gap. The authors need to conduct a proper literature review to provide up to date studies on this topic. The authors could indicate global perspective and findings of other previous studies investigating factors associated with unsafe induced abortions.

Response: Thank you for the feedback. We reviewed relevant literature of global context and have written on page 3, first and second paragraphs).

Later, the authors could mention relevant studies conducted in Nepal and the gaps needed to be addressed. The authors said that there is no study conducted at the national level, however, the authors could mention relevant studies conducted at communication level in Nepal to provide a summary of findings from previous studies.

Response: Thank you for the feedback. We reviewed relevant literature of Nepalese context and have written in full paragraph (page 4, first and second paragraphs):

Comment: The authors mentioned that ‘Some studies reported that unsafe abortion rate was higher among women with lower income, ethnic minorities, and lower education’. This looks similar to the finding of this study. Please clearly mention what are the added value of this study.

Response: Thank you for the comment, and this has been addressed in the revised manuscript (Please see last paragraph of page 4).It has been corrected.

Comment: There is no justification why the authors used 2011 and 2016 NDHS. Please mention why the authors did not use 2001 DHS or 2014 MICS (Multiple Indicator Cluster Surveys).

Response: We have not included previous surveys (NDHS 2001 and NDHS 2006) because those surveys lacked information on abortion services. In facts, those surveys have not included questions on abortion services.

Comment: What is the reference of the sentence ‘The WHO suggests that empirical research on unsafe abortion would help to re-evaluate existing programs as well as formulate appropriate strategies to improve safe abortion services.’? In addition, this sentence does not strengthen the justification of this survey because it is not an empirical research.

Responses: Thank you for the comment. References are provided for the arguments suggested. The word empirical research in this manuscript was used to reflect the practical research; and the findings from nationally representative NDHS can be the useful instrument to inform policy and practice.

Methods:

The authors should provide more details in statistical analysis.

Data source and sample:

1. I suggest the authors to describe DHS in general.

Responses: description of DHS is provided in page 5 second paragraph.

2. What is the total sample size in the end? How did you come to this final sample size?

Response: total sample size was 911. These are pooled data of NDHS 2011 AND 2016. Detailed descriptions are provided on page 5, last two paragraphs.

3. How did you handle missing data?

Response: A total of 45 missing values were excluded from the multivariate logistic regression analysis, and this has been stated in the methods section of the revised manuscript (please see line … of page …..). In addition, we have mentioned in the limitation that we could not include gestational period, an important confounder, into the adjusted regression model because of huge missing values (20%) which in case of inclusion could bias the result (please see line … of the study limitation section of the revised manuscript).

Independent variables:

For ethnicity, the authors merged some ethnic groups with small sample size into other ethnic groups and said that these groups were similar each other. However, the authors did not provide any evidence with reference on this. Moreover, the authors need to indicate the number of sample size of certain ethnic groups instead of saying ‘small size’. How about Newari origins? This is also one of the unique and major ethnic groups in Nepal.

Response: References are provided as suggested for ethnic categorization. Like other studies[4, 6], Newari ethnic group also included into Janajati ethnic group.

‘Husband education’ is mentioned twice in the Fig 1. Conceptual framework. Please remove one.

Response: This has been corrected.

Statistical analysis:

1. the authors conducted a four staged multivariate logistic regression model but they did not explain why this method is the best to achieve the goal of the study.

Response: The four-stage technique was adopted based on four-level of data that can be divided based on its proximity to the outcome [3, 7]. This has been addressed in the revised manuscript (Please last paragraph of page 8, statistical analysis subheading under methods section). This approach is also consistent with previous studies that used Nepal DHS data [3-6] .

2. The authors mentioned unadjusted odds ratios as (aOR). Did you mean adjusted odds ratios?

Responses: Yes, aOR means adjusted odds ratio. It has been corrected.

3. The authors did not mention how to choose reference groups when performing logistic regression analysis. Please explain.

Response: references group are chosen considering the possibility of a better interpretation of the findings. In the most of cases, we choose the advantaged category as reference groups.

4. There are too many independent variables and some are highly interrelated such as women’s education and literacy and women’s occupation and working status. Have the authors checked multicollinearity?

Responses: We checked multicollinearity using Variation Inflation Factor test; however, there was not find any multi-collinearity of the variables.

Results:

Descriptive characteristics of the study population:

1. Table 1:

a. It is not clear to me what chai-square means here- is it for categories under unsafe abortion? Please specify.

Responses: It is chi-square p-value obtained from cross-tabulation of each independent variables and unsafe abortion. It has been corrected in the table.

b. I suggest the authors to revise the table 1. Column percentage and row percentage are mixed so it is confusing.

Response: It has been corrected; column percentage is deleted. Now each row per cent indicates the % of unsafe abortion out of total abortion in that category.

c. The authors used the symbol, “@ and *”. Need to check whether this is in line with the PLOS ONE guideline.

Response: PLOS ONE Guideline allows those symbols; however, we have deleted in the revised manuscript.

Unsafe abortion practices in Nepal:

1. the authors mentioned that ‘Over the study period (2007-2016) in Nepal, the total and unsafe abortion rates were 36 (95% CI: 33, 38) and seven (95% CI: 6, 8) per 1000 women aged 15-49 years respectively’. However, there is no table or figure with this data. The authors should present the results with tables or figures. If not in the main manuscript, the authors could provide data in supplementing document.

Response: We have revised methods section how it was abortion rates were calculated (see page 8 under statistical subheading). Simply abortion rates are the total numbers of abortions per thousand women of reproductive age (15-49 years). It is calculated using formula total numbers of abortion (or unsafe abortions for unsafe abortion rate) divided by total numbers of women interviewed and multiplied by 1000. .

Best is give him the table as supplementary as discussed previously.

2. Saying ‘study period 2007-2016’ is confusing. Suggest revising as ‘Data from 2011-2016 NDHS’.

Response: It has been revised as suggested.

3. The authors mentioned methods of unsafe induced abortions (medical, surgical, etc.), however, there is no data in the table. The authors should present all mentioned data in the table or figure.

Response: It has been corrected as suggested. This study aimed to identify factors associated with unsafe abortion, and putting this information in the table does not suit this study. Therefore we used this in the textual form for a general overview.

Factors associated with unsafe abortion in Nepal:

1. Table 2

a. Why there are three empty rows under predisposing factors?

Response: it has been corrected.

b. It is not easy to understand the table 2. The authors can consider presenting the results with figure to have a better visualization of results.

Response: We think the way we have presented the results in the table is good for lay health workers as well as general readers. Additionally, p values of adjusted odds ratio were made bold which indicate significance in the final model at p<0.05.

2. Even though the authors indicated that they conducted a four staged multivariate logistic regression model, there is no results of model 1-4. What are the results?

Response: Putting results from 4 stages in the paper looked very busy. We decided to use the final model for readers to navigate easily; and this has been done previously [8].

3. Also, there is no results regarding this sentence ‘To avoid any statistical bias, the results from the staged model were also checked by: (1) entering only potential risk factors with p-value < 0.20 obtained in the univariate analysis for backward elimination process, and (2) testing the backward elimination method by including all potential risk factors’. It is not clearly mentioned.

Response: We employed alternative regressions technique to confirm/validate the estimates but we found staged regression technique have provided precise estimates than other techniques.

Discussion:

In general, the authors should provide a greater explanation of the findings in the discussion section. For instance, it is not clear the implication of the sentence ‘In Mexico, the legal status of abortion varied by state; Mexico City offers abortion up to 13 weeks gestation, whereas in Brazil abortion is legal if pregnancies result from rape or incest or if the life of the pregnant woman or fetus is at risk’. Why is it important and what needs to be done to improve the situation?

Response: Thank you for your suggestions. We provided those statement to support our findings of socioeconomically poor women have higher odds of unsafe abortion. Poor Mexican and Brazilian women have also had a higher unsafe abortion because of legal barriers in those countries.

Implication:

It would be good to provide relevant references regarding the arguments of the authors. For instance, is there any studies supporting the sentence ‘From the demand side perspective, the community needs to be informed and sensitized about the use of safe abortion services. Moreover, the integration of awareness raising interventions in existing health programs could increase the demand for safe abortion services.’? It may help make the argument strong.

Response: many thank you for your suggestions. We have provided relevant references on our important arguments, including the above statements.

Reviewer #5: Abstract: Can be made more concise

Response: it has been revised as suggested.

Methods: How was the wealth index calculated? What was the assessment tool used by the NDHS?

Response: In NDHS, wealth quintiles were calculated using principal component analysis of 40 households’ asset items. In this research, household wealth quintile were categorised into three groups: the bottom 40% was referred to as poor households, the next 40% as the middle households and the top 20% as rich households.

How was the rates calculated for abortion rates and unsafe abortion rates?

Response: We have revised methods section how it was abortion rates were calculated (last paragraph of page 8, statistical analysis subheading). Simply abortion rates are the total numbers of abortions per thousand women of reproductive age (15-49 years). It is calculated using formula total numbers of abortion (or unsafe abortions for unsafe abortion rate) divided by total numbers of women interviewed and multiplied by 1000.

How was it ensured that the health practitioners who did the abortion were certified for it?

Response: In Nepal, nurses and doctors get training for abortion services, so we have included doctors and nurse as skilled providers assuming there were certified providers.

Results: Redesign the table 1 and 2. Make it clearer.

Response: tables have been revised and made more readable.

Check the numbers, there are discrepancies. If it is a case of missing data, justify

Response: It has been checked and corrected if needed.

Limitation and biases have to be mentioned

Response: it has been revised as suggested. A full paragraph has been developed for limitations and strengths.

Discussion needs some more papers which could be more contextual in the countries setting.

Responses: it has been revised as suggested. Many thank you for your insightful comments.

1. Mohan, D., et al., Analysis of dropout across the continuum of maternal health care in Tanzania: findings from a cross-sectional household survey. 2017. 32(6): p. 791-799.

2. Joshi, C., et al., Factors associated with the use and quality of antenatal care in Nepal: a population-based study using the demographic and health survey data. 2014. 14(1): p. 94.

3. Khanal, V., et al., Under-utilization of antenatal care services in Timor-Leste: results from Demographic and Health Survey 2009–2010. 2015. 15(1): p. 211.

4. Poudel, S., et al., Trends and factors associated with pregnancies among adolescent women in Nepal: Pooled analysis of Nepal Demographic and Health Surveys (2006, 2011 and 2016). 2018. 13(8): p. e0202107.

5. Akombi, B.J., et al., Child malnutrition in sub-Saharan Africa: A meta-analysis of demographic and health surveys (2006-2016). 2017. 12(5): p. e0177338.

6. Ghimire, P.R., et al., Factors associated with perinatal mortality in Nepal: evidence from Nepal demographic and health survey 2001–2016. 2019. 19(1): p. 88.

7. Victora, C.G., et al., The role of conceptual frameworks in epidemiological analysis: a hierarchical approach. 1997. 26(1): p. 224-227.

8. Ghimire, P.R., et al., Socio-economic predictors of stillbirths in Nepal (2001-2011). PloS one, 2017. 12(7): p. e0181332.

20 Sep 2019

Factors Associated with Unsafe Abortion Practices in Nepal: Pooled Analysis of the 2011 and 2016 Nepal Demographic and Health Surveys

PONE-D-19-14863R1

Dear Ms. Khatri,

We are pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it complies with all outstanding technical requirements.

Within one week, you will receive an e-mail containing information on the amendments required prior to publication. When all required modifications have been addressed, you will receive a formal acceptance letter and your manuscript will proceed to our production department and be scheduled for publication.

Shortly after the formal acceptance letter is sent, an invoice for payment will follow. To ensure an efficient production and billing process, please log into Editorial Manager at https://www.editorialmanager.com/pone/, click the "Update My Information" link at the top of the page, and update your user information. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, you must inform our press team as soon as possible and no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

With kind regards,

Russell Kabir, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

30 Sep 2019

PONE-D-19-14863R1

Factors Associated with Unsafe Abortion Practices in Nepal: Pooled Analysis of the 2011 and 2016 Nepal Demographic and Health Surveys

Dear Dr. Khatri:

I am pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximize its impact. If they will be preparing press materials for this manuscript, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

For any other questions or concerns, please email plosone@plos.org.

Thank you for submitting your work to PLOS ONE.

With kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Russell Kabir

Academic Editor

PLOS ONE