Article Text

Download PDFPDF

Falling sex ratios and emerging evidence of sex-selective abortion in Nepal: evidence from nationally representative survey data
  1. Melanie Dawn Frost1,
  2. Mahesh Puri2,
  3. Peter Richard Andrew Hinde3
  1. 1Independent researcher, Southampton, Hants, UK
  2. 2Center for Research on Environment, Population and Health Activities, Kathmandu, Nepal
  3. 3Social Statistics and Demography, University of Southampton, Southampton, UK
  1. Correspondence to Dr Melanie Dawn Frost; melanie.frost{at}


Objectives To quantify trends in changing sex ratios of births before and after the legalisation of abortion in Nepal. While sex-selective abortion is common in some Asian countries, it is not clear whether the legal status of abortion is associated with the prevalence of sex-selection when sex-selection is illegal. In this context, Nepal provides an interesting case study. Abortion was legalised in 2002 and prior to that, there was no evidence of sex-selective abortion. Changes in the sex ratio at birth since legalisation would suggest an association with legalisation, even though sex-selection is expressly prohibited.

Design Analysis of data from four Demographic and Health Surveys, conducted in 1996, 2001, 2006 and 2011.

Setting Nepal.

Participants 31 842 women aged 15–49.

Main outcome measure Conditional sex ratios (CSRs) were calculated, specifically the CSR for second-born children where the first-born was female. This CSR is where the evidence of sex-selective abortion will be most visible. CSRs were looked at over time to assess the impact of legalisation as well as for population sub-groups in order to identify characteristics of women using sex-selection.

Results From 2007 to 2010, the CSR for second-order births where the first-born was a girl was found to be 742 girls per 1000 boys (95% CI 599 to 913). Prior to legalisation of abortion (1998–2000), the same CSR was 1021 (906–1150). After legalisation, it dropped most among educated and richer women, especially in urban areas. Just 325 girls were born for every 1000 boys among the richest urban women.

Conclusions The fall in CSRs witnessed post-legalisation indicates that sex-selective abortion is becoming more common. This change is very likely driven by both supply and demand factors. Falling fertility has intensified the need to bear a son sooner, while legal abortion services have reduced the costs and risks associated with obtaining an abortion.

  • Public Health

This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: and

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Article summary

Article focus

  • Little evidence exists concerning whether or not the legal status of abortion is related to the incidence of prenatal sex selection; Nepal provides an interesting case study.

  • Sex ratios at birth in Nepal before and after abortion was legalised in 2002 are calculated in order to test for evidence of sex-selective abortions.

  • This study seeks to identify characteristics of women experiencing the lowest sex ratios and thus those who are most likely to have a sex-selective abortion.

Key messages

  • There is clear evidence of a substantial increase in sex selection since 2002, even though this remains strictly prohibited.

  • The sex ratio for second and third-born children where previous births were females decreased substantially in almost all regions of Nepal, indicating the increased usage of sex-selective abortions.

  • Women who are wealthier, better educated and living in urban areas are most likely to use prenatal sex selection.

Strengths and limitations of this study

  • This is the first study to show that sex-selective abortions are becoming increasingly common in Nepal at the national level.

  • The survey data used did not allow for a more detailed geographical breakdown, which would have been useful for policymakers.


Selective abortion of females has been found to occur extensively in several Asian countries. Estimates suggest that between 3.1 and 6.0 million sex-selective abortions took place in India in the 2000s.1 China has a well-documented history of so-called ‘missing girls’ resulting from a combination of the one child policy and preference for sons, with as few as 847–877 girls born for every 1000 boys.2 ,3 Both China and India have well-established abortion services provided by national and regional governments; sex-selective abortion is illegal in both countries.4 ,5

An interesting question is the extent to which state provision of abortion services is associated with the selective abortion of females. Assessing this using data from China and India is difficult as the provision of abortion services has increased gradually.4 ,6 However, the neighbouring country of Nepal saw an abrupt change between 2002 and 2004, which provides the opportunity to measure the impact of state provision of abortion services on sex-selective behaviour. Abortion was legalised in Nepal in 2002 and Comprehensive Abortion Care (CAC) services were provided by the government from 2004 onwards.7 Sex determination is expressly prohibited and anyone found performing (or facilitating) an abortion on this basis can be punished with 1-year imprisonment, but otherwise the law is now liberal. Prior to 2002, unsafe abortion was the third largest cause of maternal mortality and it was not uncommon for women living in the south of the country to visit India if they wanted an abortion, though the scale of this is not known.7 Studies of post-abortion care at 10 major hospitals showed that between 20% and 60% of the women admitted as obstetric and gynaecological patients were abortion complication cases.8

Nepal has revealed high levels of preference for sons since the World Fertility Surveys first documented the phenomenon in the 1980s.9 Daughters are often considered to be an economic burden because of the dowry system and the fact that a girl will join her husband's family after marriage. Sons are highly prized because they continue the family name, are deemed crucial to perform funeral rites, are likely to bring in a daughter-in-law into the family and are expected to provide support for their parents in old age.10–12

Prior to 2002, preference for sons in Nepal was mainly evidenced through differential stopping behaviour (ie, the decision to have another child depended upon the sex composition of previous children). It has been estimated that such behaviour caused the fertility rate to be at least 6% higher and contraceptive use to be as much as 24% lower than it would otherwise have been.9 ,10 It therefore seems possible that, despite the strict prohibition of abortion on the grounds of the sex of the child, Nepalese will have taken advantage of the opportunities provided by the legalisation of abortion to influence the sex composition of their children.

Previous research suggests that sex-selective abortions occur in Nepal, but the evidence is either qualitative or based on small scale data.13 A study in Kathmandu recorded the sex and parity of all babies born in a single hospital during a 5-year period (2003–2007) and found that, while for first-born children the sex ratio was relatively normal at 943 females per 1000 males, the sex ratio for second order births was 847 and for higher order births it was 565.14 Unfortunately, the sample sizes were too small to draw any significant conclusions. A rapid assessment of sex ratios at birth in 1991 and 2001 consistently found more male than female babies being born in a number of districts of Nepal bordering India.15 A study conducted in 2007 revealed that legal sanctions against prenatal sex determination and sex-selective abortion had not stopped medical practitioners and sonographers in Nepal from providing such services.16

Abnormal sex ratios can be the result of mechanisms other than sex-selective abortion: one example is through Hepatitis B, which can cause the sex ratio to be skewed towards males, though this is somewhat controversial17; there are also other biological explanations, but none of these would operate differentially across birth orders. If sex ratios vary substantially across birth orders, then the explanation for this is almost certainly the use of sex-selective abortion. In India, for example, recent research indicates that the sex ratio of first-born children has not changed significantly since 1990, while the sex ratio of second-born children where the first was female was as low as 786 in the year 2000 and has not been above 900 since 1992.1 Research from China has also found that sex ratios differ by parity—in 2004–2005, the sex ratio of first-born children was 925, while for second-born children it was 699.3

It should be noted that even when sex-selective abortion is occurring, changes in its prevalence can occur for reasons related to the demand for, rather than the supply of, abortion services. Where fertility is declining rapidly, the preference for sons may be intensified since the overall desired number of children will fall faster than the desire for sons, and pressure on couples to balance the sex composition of their children may contribute to further demand for prenatal sex selection at low parities.12 ,18 Nepal has seen one of the most dramatic fertility declines in history, with the total fertility rate falling from 4.1 to 2.6 in just 10 years.19

The aim of this paper, therefore, is to examine sex ratios of births prior to and after the introduction of Comprehensive Abortion Care in 2004 to assess whether, given the religious and socioeconomic value given to sons, the legalisation of abortion and the availability of prenatal sex-determination technologies and abortion clinics were associated with a change in the proportion of male babies.



Over 40 000 women were interviewed in four rounds of the Nepal Demographic and Health Survey (NDHS), conducted in 1996, 2001, 2006 and 2011. All four surveys were nationally representative and used a two-stage stratified sampling scheme. All women were asked about their fertility and full birth histories were obtained from the 31 842 women aged 15–49; these birth histories included information on over 111 000 births. Full details of the NDHSs can be found elsewhere.19–22

A module on induced abortion was included in the 2011 NDHS for the first time. This included questions to ascertain the prevalence of abortion as well as the reasons for it. Overall, 7.5% of pregnancies were reported to have ended in abortion in the 5 years prior to the survey. This varied substantially depending on the woman's socioeconomic characteristics, with 17.5% of pregnancies ending in abortion for women in the highest wealth quintile. Of the 506 women who gave a reason for their abortion, 3.16% reported that it was owing to the child's sex, but the subject area tends to have a high incidence of misreporting, so indirect methods are likely to be the best method of studying the phenomenon.23

Statistical analysis

In the absence of intervention, the sex ratio at birth in human populations is essentially a biological constant with relatively little variation. For every 1000 males born, there will generally be between 950 and 975 females.24 ,25 This should not vary by birth order.

To control for the fact that demand for sex-selective abortions will vary by birth order and sex of previous children, conditional sex ratios (CSRs) are used in our analysis, specifically the CSR for second-born children where the first-born was female. This sex ratio is generally held to be existent where the evidence of sex-selective abortion will be most visible, since the motivation for sex-selection will probably be at its highest; it has been used frequently to provide evidence for the widespread use of sex-selective abortion in India.1 ,23 If a woman is sufficiently motivated (or pressured) to have a son that she is prepared to have an abortion to achieve that end, it is unlikely that she will wait until a much higher parity than two or three.

The sex ratios presented in this paper are the number of female births per 1000 male births, calculated as (Pf/Pm)×1000, where Pf is the proportion of total female births and Pm is the proportion of total male births. The sex ratios are weighted to account for the sampling scheme used in NDHS. We only included births reported to have happened within 10 years of the survey date, since extremely long recall periods tend to be less accurate, and misreporting may differ by sex of the baby. For time periods where more than one survey was available, a weighted average of the different surveys was used.

Wealth tertiles were produced using principal components analysis; this was performed separately for urban and rural areas in order to enable us to look at the relative contribution of women at different wealth levels in diverse environments. Tertiles were used, rather than the more traditional quintiles, owing to sample size considerations.

All analyses were conducted in STATA (V.12.0).


Sex ratios at birth were found to vary substantially by parity, the sex of the previous offspring and the time period during which the births occurred. Table 1 shows the trends in different conditional sex ratios over the 18-year period of 1992–2010. Years were grouped as the sample sizes for single years were small. The overall sex ratio is never significantly different from the expected range of 950–975 and does not show a significant change over time; the same is true of the sex ratio of firstborn children. However, for second-born children where the first-born was female, the sex ratio shows a substantial decline after the legalisation of abortion. In 2001–2003, it was 947; in 2004–2006 (when government abortion services became available in most district headquarters and urban and semi-urban areas), it dropped to 830 and by 2007–2010, it had fallen further to just 742, despite the overall sex ratio being 933. For third-born children, where the first two children had both been female, the sex ratio was also very low at 767, compared to 892 if the first two children had both been male, or 935 if one was male and the other was female. The sex ratio for second-born children in 2007–2010 was low even for mothers whose first-born was male at 822. Of the 4521 births recorded in NDHS 2011 during 2007–2010, 1247 were second order births and just 559 (44.8%) of these were female. Assuming the biologically expected sex ratio at birth, we would have expected at least around 650 females to have been born, given the number of males.

Table 1

Conditional sex ratios at birth (females per 1000 males), by birth order, 1992–2010

It should be noted that in 2007–2010, only 14% of all births in Nepal were second order after a firstborn girl. This explains why the overall sex ratio at birth is not greatly affected even by substantial changes in the sex ratio for this subset of births.

The sex ratio for second-born children where the first was female is thought to be the strongest evidence available for sex-selective abortion, short of direct data. If this particular CSR also varies by socioeconomic characteristics, then this is even stronger evidence that sex-selective abortion is being used, since knowledge of, access to and uptake of abortion will not be equal in all parts of society. Figure 1 shows how the CSR has varied over 4 NDHSs by mother's education. The sex ratios shown are those related to the 5 years prior to each survey. Table 2 shows the CSR and overall sex ratio prior to and post-2004 (when CAC services first became available), the percentage change in the CSR and the significance of the fall using a one-sided t test. A dramatic decline in the sex ratio is evident for those women who have at least a secondary leaving certificate or above. In the 5 years prior to NDHS 2006, the CSR for this group of women was just 653 females for every 1000 males, and NDHS 2011 showed a CSR of just 368, while the CSR was almost 50% lower after 2004 (Figure 2). The fall in the CSR after 2004 was significant at the 5% level for all educational groups except those with no education.

Table 2

Conditional sex ratios of second order births where the first-born was female and overall sex ratios, prior to and post-introduction of CAC services

Figure 1

Conditional sex ratio of second order births where the first-born was female, by educational level of the mother, over time.

Figure 2

Conditional sex ratio of second order births where first born was female, by wealth tertile of the household in urban and rural areas, over time.

Figure 2 shows the same CSR split by wealth tertile in urban and rural areas for the five year period before each NDHS. In the 2011 NDHS rural women in all wealth tertiles display lower sex ratios than would be expected, with the sex ratio in all wealth groups below 800. A wealth gradient is visible, with the richest women being least likely to have a girl, but it seems that even poorer women are finding ways of manipulating the sex composition of their children. For urban women in the 2011 NDHS, the wealth gradient is dramatic, with the poorest urban women having a higher sex ratio than rural women, but the richest urban women exhibiting the lowest sex ratio of all at just 326, indicating where sex-selective abortion is likely to be most prevalent. The fact that the CSR is lowest among richer urban women is unsurprising, as they would be expected to be able to access abortion services most easily and have the greatest knowledge about the availability of abortion services; that said, the extent of the shortfall in female births is remarkable. Table 2 shows that the fall in the sex ratio was significant for all wealth groups in rural areas, but only for the richest in urban areas. Nonetheless, the CSR post-2004 was substantially less than 900 for all wealth groups of urban women.

There were regional differences in the CSR and how much it changed after the legalisation of abortion. Table 2 shows the conditional sex ratio for 13 subregions of Nepal for the 10 years prior to 2004 and the period post-2004. The regions are defined on the basis of altitude (‘mountain’, ‘hill’ and ‘Terai’ or plains) and longitude (‘far western’, ‘western’, ‘mid-western’, ‘central’ and ‘eastern’). The CSR declines between the two periods in every region but one (the Eastern Mountain, where the CSR prior to 2004 was below 800). Two regions show declines which are significant at the 5% level, and six regions (accounting for around two-thirds of the population of Nepal) show declines which are significant at the 10% level. The most substantial decline was visible in the Central Terai where the CSR fell by almost 30% from 989 to 701, though both the Mid-Western Hill and the Eastern Terai had lower sex ratios after 2004. Prior to 2004, just one region had a CSR below 800 and 11 out of 13 had CSRs of over 900. In the period after abortion services became available, the CSRs of 10 regions dropped below 900, with six falling below 800 and one below 700.


These findings suggest that since abortion was legalised in Nepal, sex-selective abortion has become more common. We have found no evidence that the natural sex ratio of births (unaffected by human intervention) is outside the normal range for human populations; in particular, the sex ratio of a firstborn does not differ significantly from the expected level of 950–975 females per 1000 males. However, for second-born children, especially where the firstborn was female, there has been a sharp (and significant) fall in the number of girls compared with boys. The sex ratio of third-born children where the first two were females has also fallen substantially in recent years, while for children of higher parities the sex ratio remains within the expected range; this may indicate a tendency for those women having more than four children not to use contraception or abortion services.26 Sex ratios of births are lowest for richer, more educated women, especially in urban areas, and differ substantially by parity and the sex composition of previous children.

The data we have used were collected over a course of 16 years and include over 40 000 Nepali women, but are still not sufficient to allow the identification of all the sub-populations experiencing low sex ratios. The regional estimates of sex ratios were limited to relatively large areas. Census data may yet prove fruitful for achieving a finer level of aggregation. The 2001 census indicated that no district had an abnormal population sex ratio for infants. It will be interesting to find out if almost 10 years after abortion was legalised, the infant sex ratios in any particular districts show a decline. The survey data analysed here indicate that this may well be the case, but it should be remembered that the extremely low CSRs are masked by a relatively normal overall sex ratio at birth.

Apart from the fact that the results reported in this paper are based on a relatively small number of births, some underreporting of girls in full birth histories is a possibility. However, it seems improbable that the underreporting would be greater than average for second order and third order births, or among rich, urban or better educated women in more recent surveys.

Although we have been able to suggest that sex-selective abortion has increased in Nepal since 2004, we have not been able to enumerate the size of the problem. It is hoped that census data will allow the number of sex-selective abortions in recent years to be estimated. However, the CSR we report for second order births where the first-born was female is just 742, which is as low as the same CSRs found in India, where sex-selective abortion is known to be widespread.1 ,23 It seems likely that sex-selective abortion is becoming a more common phenomenon in Nepal, a trend which mirrors that in India and China. Recent evidence from NDHS data suggest that the proportion of pregnancies aborted increases with birth order and is higher for urban, richer and better-educated women.19

The timing and abruptness of the change in CSRs suggests that it is associated with the legalisation of abortion and the state provision of abortion services in the country. CAC services have only been available since 2004, but in some regions the CSR of second-born children where the first-born was male has fallen by well over 200 in that time. However, we do not mean to suggest a simple relationship between the introduction of CAC services and the change in CSRs. It is more likely that both demand and supply factors have been at work. The decline in fertility has increased the importance that at least one son be born among the first three children. This has intensified the demand for abortions on the grounds of the sex of the fetus. Prior to 2004, Nepali women desiring abortions had to be prepared to incur either substantial health risks (from illegal services) or costs (from going to India) or both. The effect of the introduction of CAC services in 2004 was to reduce both the risks and the costs, and hence to increase the number of abortions in Nepal. The reduction in risks/costs applied to all abortions, but since a major (and increasingly important) reason for demanding an abortion in Nepal is the sex of the fetus, there has been a rise in the number of abortions performed for this reason. It is interesting to note that the change in the CSR for second births following a firstborn girl was greatest among urban educated women, who would have been most able to afford abortions even prior to 2004. The decline in fertility has been most acute among this group; if cultural forces leading to a strong desire for at least one son had remained universal in Nepal, their demand for sex-selective abortions might have increased more rapidly than that of rural and less well-educated women.

Clearly, there are policy implications in terms of education about the long-term social and demographic consequences of sex-selective abortion. While educating medical practitioners is one important step, previous research has shown that medical practitioners believe that if a woman wishes to have a sex-selective abortion, she can easily obtain one in India; it may also be hard for practitioners to judge a woman's reasons for wanting an abortion.13 ,27 For these reasons, it is also paramount to educate wider members of the community about the illegality of sex determination tests for the sole purpose of selective abortion and the health risks associated with second trimester abortions.

Advocacy and educational campaigns among groups with particularly low sex ratios will be important; literate, urban women are one target group. Those living in the Terai with easy access to the Indian border are another. It is also important to remember that the use of sex-selective abortion has risen not only because of sociocultural preferences for sons but also because girls are seen as an economic burden. Ultimately, combating sex-selective abortion will require the empowerment of women throughout Nepali society. Programmes benefiting female children would be a start; these might include maternity incentives to mothers who give birth to a girl, education scholarships and free vocational training for girls, as well as economic incentives for the parents of girls.


This work was undertaken independently and there were no funding sources. The views expressed in this manuscript reflect those of the authors and do not necessarily reflect the views of any institution. Dr Andrew “Amos” Channon provided the wealth indices used within this paper.



  • Contributors MDF conducted the data analysis. MDF, MP and PRAH were involved in interpreting the data and writing the paper and have approved the final version. MDF is the guarantor.

  • Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The full datasets are available at Consent was not obtained for this study, but participants gave informed consent for the original survey; the presented data are anonymised and there is no risk of identification.