Open Access Journal of Contraception Dovepress

open access to scientific and medical research

Open Access Full Text Article

ORIGINAL RESEARCH

Roles of Proximate Determinants of Fertility in

Recent Fertility Decline in Ethiopia: Application of
the Revised Bongaarts Model
This article was published in the following Dove Press journal:
Open Access Journal of Contraception

Seifadin Ahmed Shallo Introduction: In Ethiopia, the fertility rate declined from 5.4 in 2005 to 4.6 by 2016. Many
factors have been contributing to this decline. Understanding the factors contributing to the
Department of Public Health, College of
Medicine and Health Sciences, Ambo fertility decline and their level of fertility inhibiting effect has a paramount policy implication
University, Ambo, Ethiopia in any country. This study aimed to assess the contribution of the four proximate determi-
nants of fertility, ie, contraception use, postpartum infecundity, marriage and abortion rate, to
fertility decline in Ethiopia since 2005.
Methods: This study used publicly available data from the Ethiopia Demographic and
Health Surveys (EDHS) of 2005, 2011 and 2016. The EDHS data were the representative
data collected from the reproductive-age women through a cross-sectional study. The revised
and fine-tuned Bongaarts model of proximate fertility determinants was used for data
analysis. The components needed for the analysis were extracted from the full EDHS data
using the STAT compiler. Finally, the analysis was done using Microsoft Excel.
Results: Of the four proximate determinants of fertility, postpartum insusceptibility con-
tributed the highest fertility inhibiting effect in all three EDHS, and its level was also more
prominent among the poorest women. While post partum infecundity, marriage and abor-
tion had a relatively constant effect on fertility over the last 15 years, the fertility inhibiting
effect of contraceptive use significantly increased from 15% to 37%.
Conclusion: In conclusion, fertility variation in Ethiopia is largely due to the three inter-
mediate determinants of fertility. Over the last one and half decades, contraceptive use was
the single most important determinant responsible for fertility decline in Ethiopia. To achieve
fertility at replacement level, the country needs a contraceptive prevalence rate of 69%, an
increment of nearly 100% from its current contraceptive prevalence rate.
Keywords: proximate determinants, Bongaarts revised model, Ethiopia

Background
It is believed that the rate of population growth implies economic growth of any
given country. The recent fertility decline in many African countries is assumed to
be one of the opportunities paving way for economic improvement. However, the
rate of fertility decline is not equivalent with other continents. The demand for
a high number of children is common in sub-Saharan Africa, and the United
Nations Population Divisions estimated the TFR of the region to be 4.76 in
2017.1,2 Studies indicate that the fertility of a given community is affected by socio-
demographic factors such as age at marriage, women’s literacy status, and contra-
Correspondence: Seifadin Ahmed Shallo
Email Seifadinahmed8226@gmail.com ceptive use.3,4

submit your manuscript | www.dovepress.com Open Access Journal of Contraception 2020:11 33–41 33
DovePress © 2020 Ahmed Shallo. This work is published and licensed by Dove Medical Press Limited. The full terms of this license are available at https://www.dovepress.com/terms.
http://doi.org/10.2147/OAJC.S251693
php and incorporate the Creative Commons Attribution – Non Commercial (unported, v3.0) License (http://creativecommons.org/licenses/by-nc/3.0/). By accessing the
work you hereby accept the Terms. Non-commercial uses of the work are permitted without any further permission from Dove Medical Press Limited, provided the work is properly attributed. For
permission for commercial use of this work, please see paragraphs 4.2 and 5 of our Terms (https://www.dovepress.com/terms.php).
Ahmed Shallo Dovepress

With the high-level efforts of governmental and non- proportion of married women; prevalence of contracep-
governmental organizations’ promotion of contraceptive tion; rate of induced abortion, frequency of sexual inter-
use6, there was an improvement in modern contraceptive course; sterility; and spontaneous intrauterine mortality
prevalence in the last three decades in many sub-Saharan and duration of the fertile period. The degree at which
African countries.5 In Ethiopia, the prevalence of modern these different intermediate factors affect fertility varies
contraceptives increased from 14.7% in 2005 to 35% in between societies.15 In the late 1970s, John Bongaarts
2016. However, with a total fertility rate of 4.6, the coun- developed the modified set of proximate determinants of
try has not achieved the TFR of 4 and contraceptive fertility containing four elements: marriage/cohabitation;
prevalence of 44% as targeted in population policy of the induced abortion; contraception; and postpartum in-
country formulated in 1993.6–8 fecundity. Bongaarts believed that these four components
In sub-Saharan Africa, nearly 50% of the women marry were most important and believed to simplify the model
at the age of 18 and about 70% marry by the age of 20. The for computing the fertility rate and the contribution of each
rate of early marriage is inversely related to the female PD. However, with a significant change in population
literacy rate. Where there is a high literacy rate, the rate of reproductive behaviors and research recommendations in
early marriage is low. According to data from EDHS in recent times, some original assumptions have become less
2016, the median age at marriage in Ethiopia was 16.6 accurate over time and necessitate modification.
years. Age at marriage affects the total fertility rate of Accordingly, the existing assumptions developed by John
a given population. This is because if a woman marries at Stover in 1998 were updated with some modifications in
an early reproductive age, she will have a long reproductive 2015 on three of the four components.14,16
age duration. In addition, women who marry at an early age
are less likely to be educated. This will result in low use of Bongaarts Proposed Revisions
contraceptive methods. In Ethiopia, the women’s literacy Marriage/Union/Sexual Exposure
rate is low. Only 17.2% of the women in the country In the previous model, it was considered that sexual activ-
attended an educational level of high school and above.7,9,10 ity and childbearing happens only among married women.
In Ethiopia, the prevalence of premarital sexual engage- But extra-marital sexual activity and childbearing are
ment is increasing markedly. Studies indicated the preva- becoming common in both developing and developed
lence of premarital sex reaches nearly 20–54% with countries. Based on this justification, Bongaarts proposed
incremental trends from time to time.11,12 In 2014, an antici- to estimate the number of women who are exposed to the
pated 620,300 abortions were performed in Ethiopia. This risk of childbearing as the sum of married women and
corresponds to an annual rate of 28 abortions per 1000 unmarried women. The name of the index was also chan-
women aged 15–49, an increase from 22 per 1000 in 2008.13 ged to the index of sexual exposure instead of an index of
Davis and Blake (1956) suggested two types of factors marriage (Cm).14,17
mainly affect fertility: the direct or proximate determinants
and indirect determinants or background factors. The prox- Contraception Prevalence
imate determinants (PD) of fertility are both biological and In the model modified by John Stover, the assumption of
behavioral determinants that affect the fertility directly. the postpartum in-fecundity period overlaps with postpar-
The indirect factors such as socio-demographic and culture tum contraceptive use was ignored. But, the recent incre-
influence fertility through these proximate determinants ment in contraceptive prevalence as a result of postpartum
but indirectly. If an intermediate fertility variable, such contraceptive promotion may result in the overlap and, if
as the prevalence of contraception changes, then fertility not taken into consideration, can significantly affect the
necessarily changes (assuming the other intermediate fer- model.17 In such cases, excluding the overlapping period
tility variables remain constant), while this is not necessa- should be considered. In addition, since the contraceptive
rily the case for an indirect determinant such as income or prevalence varies with age, index of contraception should
education. If measured and modeled appropriately, PD can consider age-specific PD models rather than the aggregate
express the variability in fertility with relatively less or approach.14 In addition to what Bongaarts has proposed,
a few errors.14,15 the total contraceptive prevalence rate should be consid-
In the mid-1950s Davis and Blake proposed eleven ered instead of contraceptive prevalence among only mar-
proximate determinants of fertility which include: the ried women/women in union.

submit your manuscript | www.dovepress.com Open Access Journal of Contraception 2020:11

34
DovePress
Dovepress Ahmed Shallo

Abortion probably introduced on one or more of the indices. Such

Estimates of the number of births averted by induced errors originated mainly from researchers’ mistakes rather
abortion are mainly based on numerical exercises using than the model error. Therefore, this paper also fills these
mathematical reproductive models. It is strongly influ- gaps. Furthermore, the contribution of each four PD index
enced by the practice of family planning service use fol- should also be identified as these have great policy impli-
lowing induced abortion. In the absence of contraceptive cations for the country. Consequently, the objective of this
use, induced abortion can reduce 0.4 births and with rela- paper was to assess the magnitude of the proximate deter-
tively effective contraceptive use, 0.8 births will be minants of fertility and their variation based on some
averted by induced abortion. For this concept, the follow- selected socio-demographic factors.
ing formula was developed:
Materials and Methods
b¼ 0:4ð1 þ uÞ
Study Setting and Sources of Data
Where b=is the number of births averted by induced abor- This study used publicly available data from the Ethiopian
tion and u= ideally considered to be the proportion of demographic and health surveys of 2005, 2011 and 2016. The
protected women among induced abortion women. EDHS data were representative data collected through a cross-
The fertility reduction associated with a given level of sectional study from the reproductive-age women living in the
the total abortion rate is calculated as: nine regional states of the country and the two cities, ie, Addis
A ¼ b x TA ¼ 0:4ð1 þ uÞ x TA Ababa and Dire Dawa. The data collection procedure, tools
used, ethical issue and other detail were described in each
Where TA=total abortion rate. A=the mean number of EDHS. The total number of women participated in the study
births averted per woman by the end of her reproductive were 15,683, 16,515 and 14,070 in 2016, 2011 and 2005
age.14,16 EDHS, respectively. Of these, the sexually active unmarried
The index of induced abortion is computed as the ratio women account 0.02, 0.03, and 0.01, respectively.6,7,19 The
of the observed total fertility rate, TFR, to the estimated components needed for the current Bongaarts model analysis
total fertility rate without induced abortion, TFR + A, were extracted from the full EDHS data using the STAT
TFR compiler, and also with manual extraction in case the data
Ca ¼
TFR þ bxTA not available directly from the STAT compiler. Finally, the data
But, in the revised Bongaarts model, the formula is mod- were analyzed using Microsoft excel.
ified to:
b ¼ 18:5þi
14
instead of b=0.4 (1+u) while the other for-
Data Analysis Methods
The Bongaarts model justifies the reason why fertility cannot
mula is unchanged.14
reach its potential maximum of 15.3. Bongaarts presented the
In the revised model, the number of births averted per
principal proximate factors namely: effect of contraception;
abortion was considered to be the ratio of the mean repro-
the effect of sexual exposure; the effect of induced abortion;
ductive time associated with abortion to the mean repro-
and postpartum in-fecundity/insusceptibility. These four
ductive time associated with live birth, which is estimated
components are inhibiting the fertility of a given community
to be 14 and 18.5, respectively, and added postpartum in-
from reaching its maximum theoretical fecundity rate of
fecundity duration.14
15.3. The combinations of these four components were
In summary, there was one previously published article
used to determine the fertility rate and the effect and con-
by Laelago et al on proximate determinants of fertility
tribution of each component on fertility reduction.16
applying the unrevised Bongaarts model to the Ethiopian
DHS of 2011 and 2016.18 However, it is likely that there In this study, the 2015 revised and fine-tuned
are major errors in the above paper. For instance, even Bongaarts model of proximate fertility determinant was
though it is clear that fertility in the country is decreasing used.14 Bongaarts proposed the total fertility rate (TFR)
of a given community is the product of the four indexes
over time, it was reported as if the TFR of the country
and put the formula as follows:14,16
increased from 4.04 (in 2011) to 4.14 (2016) which is not
logical. The errors observed on the cumulative TFR were TFR ¼ CmðaÞ  CcðaÞ  CiðaÞ  CaðaÞ  ff (1)

submit your manuscript | www.dovepress.com

Open Access Journal of Contraception 2020:11 35
DovePress
Ahmed Shallo Dovepress

Where TFR=total fertility rate. Cm=index of marriage, inhibition effect will be also high and vice versa. In general, the
Cc=index of contraception, Ci=index of postpartum in contraceptive index is estimated as follows:14
fecundity, Ca= index of abortion, and ff=total fecundity Cc ðaÞ ¼ 1  r ðaÞ ðu ðaÞ  o ðaÞÞ e ðaÞ (6)
rate, which is assumed to be 15.3.
The model treats each PD as a factor that inhibits
fertility. Each index has values that range from 1 to 0 Where: Cc=index of contraceptive use, u (a) =contracep-
depending on the degree of fertility inhibition. The index tive prevalence (among sexually exposed women), O (a)
equals 1 if it has no fertility inhibition effect, and zero =contraceptive use overlap with postpartum infecundabil-
when it has a 100% fertility inhibition effect. The mea- ity, e (a) =average contraceptive effectiveness, r (a)
surements of each index are discussed below in detail. =fecundity adjustment.
For the computation of the index, the prevalence of
contraceptive use among both married and unmarried
Estimation of Index of Sexual Exposure
women was used, mean contraceptive effectiveness was
(Cm) calculated for each user type of contraceptive and
This index measures the extent to which sexual exposure
weighted, and the contraceptive use overlap with postpar-
(includes formal marriage and cohabitation) is contributing to
tum insusceptibility was considered to be zero, assuming
the fertility rate of a given community. If there is a high level
that since the prevalence of contraceptive use is low in the
of sexual exposure in the community, the fertility inhibiting
country especially during the postpartum period, the effect
effect of sexual exposure will be low and vice versa.
of overlap is considered to be nil. The r (a) has already
Accordingly, the index is calculated as follows:14
estimated (1.08) for some countries by Bongaarts.14
Cm ðaÞ ¼ Cm ðaÞ x wm ðaÞ (2) The contraceptive use effectiveness rates were obtained
from previous studies by Trussell and Bongaarts as follows:
fmðaÞ sterilization (0.99), oral pill (0.91), Copper-based IUD (0.99),
WmðaÞ¼ (3)
fmðaÞ injectable (0.94), implants (0.99), male condom (0.82),
Where Cm (a) is the index of marriage, wm (a) is weighted rhythm/periodic abstinence (0.76), withdrawal (0.78), lacta-
age-specific marital fertility rate, and fm is the marital tional amenorrhea and folk method (0.70).20,21 To find the
fertility rate. In a case where the age disaggregate data mean contraceptive effectiveness rates, the proportion of
are not available, Bongaarts proposed the following model women using a given method was multiplied by that specific
as a proxy measure for the marriage index:14,16 method effectiveness. Finally, the weighted mean of the effec-
tiveness was used.
Cm ðaÞ ¼ m ðaÞ þ ex ðaÞ (4)

Or Estimating Index of Postpartum

Cm ðaÞ ¼ TFR=TMFR (5) infecundity (Ci)
This index estimates the fertility inhibition effect of post-
partum infecundity due to lactational amenorrhea or post-
Where m (a) = proportion married/in union and ex (a)
partum sexual abstinence. In the absence of lactation, the
=proportion of extramarital sexual exposure, TFR=total
infecund interval immediately after childbirth is on aver-
fertility rate, and TMFR= total marital fertility rate. In
age about 1.5 months. The mean waiting time from the
this paper, I compare all of the above and find a slight
menses resume to conception to be 7.5 and the time added
difference (max error of 0.01–0.04) in estimating the sex-
by intrauterine mortality equals approximately 2 months
ual exposure index.
per birth interval. Without lactation, a typical mean birth
interval can therefore be estimated to equal 1.5 + 7.5 + 2 +
Estimation of the Index of Contraception 9 (pregnancy period) =20 months, and with lactation it
Use (Cc) equals the mean total duration of the in-fecundity period
This index measures the fertility inhibition effect of contra- (i) plus 18.5 months (7.5 + 2 + 9). The ratio of the mean
ceptive use, and it is the function of contraceptive prevalence birth intervals without and with lactation is called the
and the effectiveness of each method used. If there is a high index of lactational in-fecundity and calculated as
prevalence of contraceptives in a given community, the fertility follows:14,16

submit your manuscript | www.dovepress.com Open Access Journal of Contraception 2020:11

36
DovePress
Dovepress Ahmed Shallo

Ci ¼ Ci  ðaÞwiðaÞ  Ci (7) The value of 14 is the mean reproductive duration

expected following abortion, 18.5+ i(a) is the average
20 reproductive duration expected following live births, and
Ci ¼ (8)
18:5 þ iðaÞ i(a) represents the mean postpartum in-fecundity
interval.14,16 For this model, all components were obtained
Where i= average total duration of postpartum in-fecundity
from EDHS except the abortion rate which was obtained
due to lactation or abstinence. It was estimated as the median
from the previous study.13
duration of postpartum insusceptibility as extracted for EDHS
data.
Results
Preliminary Data
Index of Induced Abortion (Ca) Some data needed for calculating the four PD indexes
This index estimates the number of births averted by were extracted and put in Table 1.
abortion and calculated using the formula:14
TFR The Estimated Effect of the Proximate
Ca ¼ (9)
ðTFR þ b  abðaÞÞ Determinants (PD)
The detail of all four indexes with their respective years is
TAR ¼ abðaÞ (10) presented in Table 2. In this analysis, marriage delay/sex-
ual non exposure inhibited fertility by 35% (Cm=0.65) and
b  ¼ 14=ð18:5 þ iðaÞ (11) 37% (Cm=0.63), contraceptive use inhibited fertility by
29% (Cc=0.71), and by 37% (Cc=0.63) in 2011 and
2016, respectively.
Where TFR =total fertility rate, b=births averted by
Overall, the postpartum in-fecundity had the highest
induced abortion, ab(a)= abortion rate.
(42%) fertility inhibiting effect followed by contraceptive
use (reduced fertility by 37%) in 2016. Even though the
Table 1 Some Selected Reproductive Indicators from EDHS, degree of fertility inhibition varies, postpartum in-
2005–2016, Ethiopia
fecundity, contraceptive use, and delay in marriage/sexual
Reproductive Indicators 2005 2011 2016 exposure were ranked first, second and third respectively
Total fertility rate 5.4 4.8 4.6 in the order of fertility inhibition effects.
Proportion of contraceptive Use=u(a) 14.7 28.6 35.9 In 2011 the fertility inhibiting effect of delay in mar-
The total marital fertility rate 8.5 7.9 7.7 riage/sexual activity was higher than that of contraceptive
Average contraceptive effectiveness 0.91 0.94 0.95 use which was not uncommon in the country where
The median duration of postpartum 16.7 16.6 16
contraceptive prevalence is a low and unmet need for
insusceptiblity=i(a)
family planning is high. But, the effect of contraceptive
Total abortion rate 0.022 0.022 0.028 use on fertility reduction was increased from what it was
6 19 7
Note: Data from EDHS, 2005, 2011, and 2016. in 2005 (15%) by 2016 (37%).

Table 2 Estimated Index of Proximate Determinants of Fertility and Their Fertility Reduction Effect in EDHS Data of 2005, 2011 and
2016
EDHS Index of Effect on Index of Effect on Index of Effect on Index of Effect on
(Years) Sexual Fertility Postpartum Fertility Contraceptive Fertility Induced Fertility
Exposure Reduction Insusceptibility Reduction (Cc) Reduction Abortion Reduction
(Cm) (Ci) (Ca)

20167 0.65 35% 0.58 42% 0.63 37% 0.997 0.3%

201119 0.63 37% 0.57 43% 0.71 29% 0.997 0.3%
20056 0.64 36% 0.57 43% 0.85 15% 0.997 0.3%

submit your manuscript | www.dovepress.com

Open Access Journal of Contraception 2020:11 37
DovePress
Ahmed Shallo Dovepress

Fertility Differences by Some Selected Since the disaggregated data on abortion rate for different
population background was not available, country-level TAR
Backgrounds
was used in this paper, and the fertility reducing effect of
Since the most recent trends imply future planning, the
abortion is almost similar among all segment of the population
analysis of the PD with different backgrounds was confined
(Table 3).
to the data extracted from EDHS 2016. As expected, there
were fertility differences according to women’s level of
education. As the level of education increase, the fertility
Estimating the Total Fertility Rate
According to Bongaarts, variation in fertility rate among
rate decreases. This could be because the women who stay
the population is due to the four factors namely: the
in school marry later in life resulting in low fertility duration.
proportion of sexual exposure; rate of induced abortion;
Both indices of sexual exposure (Cm=0.95, 5%), and contra-
duration of postpartum in-fecundity; and contraception use
ceptive use (Cc=0.68, 32%) contributed to low fertility
prevalence and it is a product of these four PD indexes and
inhibiting effect among women with no education. In addi-
fecundity rate. The other remaining intermediate variables,
tion, the index of marriage/sexual delay also contributes less
which are considered less important, are represented in the
among women who attended higher education compared
model by the total fecundity rate (TF), which has values
with women who had ever attended primary and/or second-
around 15.3 births per woman (Table 4).
ary education. This could be because women who attended
higher education are more likely to join marriage immedi- TFR = Cm *Cc *Ci *Ca *TF.
ately after the completion of their education. Once they
become married they will start to bear children as the desire Accordingly, the estimated total fertility rate was 3.6,
to have children will be higher during this stage. 3.9, and 4.2 in 2016, 2011 and 2005, respectively. All the
The differences in PD indexes were also observed estimated TFR as per the Bongaarts model were different
among rural and urban residents. Sexual exposure has low from those observed in the EDHS report. These observed
fertility inhibiting effect among rural women, while post- variations could be mainly due to the following reasons:
partum in-fecundity has a higher fertility inhibiting effect
among rural women. This could be because urban women 1. There could be errors/variation during measuring
are less likely to breastfeed for a long duration in contrast to the intermediate variables (proximate determinates
rural women. Fertility reducing effect of contraception was of fertility) in EDHS.
highest among women in the fourth quintiles of wealth index 2. The total fecundity assumed was 15.3, which is the
and lowest among women of lowest wealth quintiles. approximation. The TF, in general, is in the interval

Table 3 Estimation of PD Index in Relation to Some Background

Background Cm Ci Cc Ca Fecund (TF) TFR-est TFR-obs Error

Residence
Urban 0.67 0.77 0.47 0.99 15.3 3.7 2.3 1.4
Rural 0.795 0.57 0.67 0.997 15.3 4.6 5.2 0.6

Educational background
No education 0.95 0.56 0.68 0.998 15.3 5.6 5.7 0.1
Primary 0.64 0.60 0.62 0.997 15.3 3.6 4.2 0.6
Secondary 0.47 0.78 0.47 0.993 15.3 2.6 2.2 0.4
Higher 0.65 0.80 0.44 0.992 15.3 3.5 1.9 1.6

Wealth quintiles
Lowest 0.86 0.56 0.8 0.998 15.3 5.8 6.4 0.6
Second 0.82 0.58 0.67 0.998 15.3 4.9 5.6 0.7
Middle 0.79 0.56 0.62 0.997 15.3 4.2 4.9 0.7
Fourth 0.74 0.65 0.58 0.997 15.3 4.2 4.3 0.1
Highest 0.67 0.76 0.49 0.997 15.3 3.8 2.6 1.2
Note: Data from EDHS 2016.7

submit your manuscript | www.dovepress.com Open Access Journal of Contraception 2020:11

38
DovePress
Dovepress Ahmed Shallo

Table 4 Estimated Total Fertility Rate Using Bongaarts Revised Model EDHS 2005, 2011 and 2016
EDHS (Years) Cm Ci Cc Ca Fecund (TF) TFR Estimated TFR-Observed

20167 0.652 0.58 0.63 0.997 15.3 3.6 4.6

201119 0.634 0.57 0.71 0.997 15.3 3.9 4.8
20056 0.64 0.56 0.85 0.997 15.3 4.6 5.4

of 13 to 17. But, if we take the TF of 17, these TFR2 1  1:08ðU2Þe

¼
variations become very low and so this may be the TFR1 1  1:08ðU2Þe
reason for the deviation from the observed total
From the above formula, we can drive the contraceptive
fertility rate in the model.20
prevalence needed as in the future (u2) as follows:
3. This difference may also have contributed by the
modification done to the model. 1
U2 ¼ x½1  TFR2=TFR1 x Cc1
1:08e2
In the fertility transition phase, Bongaarts proposed that
TFR2 ¼ 2:1; TFR1 ¼ 4:6; Cc1 ¼ 0:63 and e2 ¼ 0:95
there are four phases of fertility transition. In Phase I (TFR≥6),
fertility transition is near natural fertility and if the fertility rate Accordingly, Ethiopia needs a contraceptive prevalence of
is in Phase IV (<3), it has completed fertility transition. In 69%, which is almost double of the current contraceptive
Ethiopia, the fertility transition is in Phase III (3≤TFR≤4.5), prevalence rate, to achieve the fertility replacement level.
which is nearing to completion of fertility transition.
Discussion
Future Projection of Fertility and In this paper, the fertility inhibition effects of the four
proximate determinants of fertility were assessed using
Contraceptive Prevalence
the recently modified Bongaarts model. Compared with
From the Bongaarts model, we can project fertility in
each other, postpartum in-fecundity contributes to the
a certain period in the future or the contraceptive preva-
highest fertility inhibiting effect followed by contraceptive
lence needed to reach on a certain fertility level. To do so,
use. There were some probable justifications for this find-
some assumptions and minor modifications are required.
ing. One of the reasons could be that most women who
Such projection will help planners.
participated in the study were from a rural setting where
Assuming that we want to decide the fertility rate in the
future. From the above analysis, of the four PD, except CPR was low. Studies on the determinants of fertility in
contraceptive index, three of them ie index of sexual Sudan also reported a similar finding.22 The rate of abor-
exposure, postpartum in-fecundity and abortion index did tion has the least effect on all three EDHS. The effect of
not change significantly between the three DHS. contraceptives significantly increased by 2016 from what
Therefore, if we want to forecast what proportion of con- it was in 2005. This change could probably be attributed to
traceptives is needed to achieve the replacement fertility the increment in CPR of the country from 14.7% in 2005
level (TFR=2.1), we can do so with the following process. to 36% in 2016.
Let TFR1 be the current fertility rate, TFR2 be the The fertility inhibiting effect of the PD varies according
fertility replacement level we are intended to achieve, to some backgrounds such as level of education, residence
and Cc1 be the current contraceptive index and Cc2 be and wealth quintiles. The variation in fertility between
a contraceptive index in the future. Also, let u1 be the urban and rural is mainly contributed to delay in sexual
current contraceptive prevalence and u2 be the contracep- exposure/marriage and high contraceptive prevalence in
tive prevalence we are interested to achieve the fertility urban compared to rural. The same finding was reported
replacement level. in Zambia.23 The fertility rate difference observed among
Accordingly: women of different educational backgrounds also contrib-
uted to the fact that more educated women were also more
TFR2=TFR1 ¼ Cc2=Cc1
likely to use contraceptives and delay marriage.5,24 Of all
Which is equivalent with: the PD fertility, postpartum in-fecundity contributed the

submit your manuscript | www.dovepress.com

Open Access Journal of Contraception 2020:11 39
DovePress
Ahmed Shallo Dovepress

highest fertility inhibiting effect among the poorest women Abbreviations

in 2016. Ca, index of abortion; Cc, index of contraceptive use; Ci,
By applying the fertility decomposing formula, it was index of postpartum insusceptibility; Cm, index of mar-
observed that fertility decline between 2005 and 2016 was riage; CPR, contraceptive prevalence rate; EDHS,
mainly contributed by the contraceptive use while the Ethiopian Demographic AND Health Survey; TF, Total
effect of the other indices relatively increased the fertility Fecundity; TFR, Total Fertility Rate.
from what it was in 2005. The proportion of the fertility
inhibiting effect of abortion between 2005 and 2016 was Data Sharing Statement
zero. This means the effect of abortion was constant All data generated or analyzed during this study were
throughout the decade. As indicated above, the proportion included in this published article. Also, the whole raw
of marriage increased from what it was in 2005 by 2016. data of EDHS can be accessed online from
In Ethiopia, the age at marriage is increasing; however, the STATcompiler.com.
age at which sexual exposure first occurs is becoming
earlier. A similar finding was reported in Zambia. In gen- Ethics and Consent
eral, the trends of fertility and the effects of proximate The consent to participate in the study was already assured
determinants of fertility observed in Ethiopia were almost during primary data collection in EDHS where both writ-
similar to the other Sub-Saharan African Countries. ten and verbal informed consent was obtained from each
In conclusion, over the last decades, contraceptive use respondent. The detailed ethical issues of the collected
was the single most important determinant responsible for data were mentioned in all three EDHS from 2005–2016.
fertility decline in Ethiopia. To achieve the fertility of
replacement level, the country needs a contraceptive pre- Acknowledgments
valence rate of 69%, an increment of nearly 100% from its I would like to acknowledge the Ethiopian Central
current rate. To achieve the proposed CPR rate, meeting Statistical Agency for allowing me to access the raw data
the unmet need for family planning is the key target to be and USAID organization for enabling me to compile the
focused on. This can be achieved by ensuring service data for this work.
availability and accessibility. Activities targeting control-
ling fertility, especially contraceptive availability, and
Disclosure
accessibility issues, should give due attention to the poor,
The authors report no funding for this specific work and no
rural and uneducated women. conflicts of interest.
Some of the indices in the current finding significantly
differed from the finding reported by Laelago et al,18
References
where they had applied the unrevised model. The variation
1. Beatty A (Population Readings) - Recent Fertility Trends in Sub-Saharan
was observed in postpartum in-fecundity and abortion Africa: Workshop Summary. first. BEATTYA, ed. Washington (DC): The
indices. This resulted in variation of the overall fertility National Academies Press; 2016. doi:10.17226/21857
estimation of the model which is believed could mainly 2. Recent fertility trends in Sub-Saharan Africa - NCBI Bookshelf.
Available from: https://www.ncbi.nlm.nih.gov/books/NBK350195/.
originate from the modification in the model. In addition, Accessed December 19, 2019.
in a country like Ethiopia, where there is insufficient data 3. Adhikari R. Demographic, socio-economic, and cultural factors affect-
ing fertility differentials in Nepal. BMC Pregnancy Childbirth.
on abortion rate or low prevalence of abortion, excluding 2010;10(1):19. doi:10.1186/1471-2393-10-19
the abortion index from the model does not affect the 4. Chandiok K, Mondal PR, Mahajan C, Saraswathy KN. Biological and
social determinants of fertility behavior among the jat women of
model’s fertility estimating power. Haryana state, India. J Anthropol. 2016;2016:1–6. doi:10.1155/2016/
5463168
5. Emina JBO, Chirwa T, Kandala NB. Trend in the use of modern
Limitation of the Study contraception in sub-Saharan Africa: does women’s education matter?
There could be errors/variation during primary data collec- Contraception. 2014;90(2):154–161. doi:10.1016/j.contraception.20
14.02.001
tion which can directly or indirectly affect the intermediate 6. Macro O; Agency/Ethiopia CS. Ethiopia Demographic and Health
variables (proximate determinates of fertility) in all EDHS. Survey 2005. 2006.
7. Central Statistical Agency (CSA) [Ethiopia] and ICF. 2016. Ethiopia
So, interpretation of the finding should put this into Demographic and Health Survey 2016. Addis Ababa, Ethiopia, and
consideration. Rockville, Maryland, USA: CSA and ICF.

submit your manuscript | www.dovepress.com Open Access Journal of Contraception 2020:11

40
DovePress
Dovepress Ahmed Shallo

8. Teller C, editorThe Demographic Transition and Development in 17. Stover J. Proximate determinants of fertility Google Search; 1998.
Africa. Springer Science+ Business Media BV; 2011. doi:10.1007/ Available from: https://www.google.com/search?q=John+Stover+(1998)
978-90-481-8918-2 +proximate+determinants+of+ferrtlity&oq=John+Stover+(1998)+proxi
9. Singh S, Samara R. Early marriage among women in developing mate+determinant s+ of+ferrtlity+&aqs= ch rome. 69i 57j 33.
countries. Int Fam Plan Perspect. 1996;22(4):148–157+175. 10880j1j1&sourceid=chrome¡UTF–8. Accessed December 26, 2019.
doi:10.2307/2950812 18. Tariku Laelago YH, SY. Proximate determinants of fertility in
10. Raj A, Saggurti N, Balaiah D, Silverman JG. Prevalence of child Ethiopia; an application of revised Bongaarts model. BMC Reprod
marriage and its effect on fertility and fertility-control outcomes of Heal. 2019;16(13). doi:10.1186/s12978-019-0677-x
young women in India: a cross-sectional, observational study. Lancet. 19. Central Statistical Agency [Ethiopia] and ICF International. 2012.
2009;373(9678):1883–1889. doi:10.1016/S0140-6736(09)60246-4 Ethiopia Demographic and Health Survey 2011. Addis Ababa,
11. Premarital sexual practice and its predictors among university stu- Ethiopia and Calverton, Maryland, USA: Central Statistical Agency
dents: institution based cross-sectional study. Available from: https:// and ICF International.
www.panafrican-med-journal.com/content/article/28/234/full/. 20. Bongaarts J. The fertility-inhibiting effects of the intermediate ferti-
Accessed December 25, 2019. lity variables. Stud Fam Plann. 1982;13(6–7):179–189. doi:10.2307/
12. Bogale A, Seme A. Premarital sexual practices and its predictors 1965445
among in-school youths of shendi town, west Gojjam zone, North 21. Trussell J. Contraceptive failure in the United States. Contraception.
Western Ethiopia. Reprod Health. 2014;11(1):49. doi:10.1186/1742- 2004;70(2):89–96. doi:10.1016/j.contraception.2011.01.021
4755-11-49 22. Fertility Determinan in Sudan. Analysis of multiple indicator cluster
13. Moore AM, Gebrehiwot Y, Fetters T, et al. The estimated incidence survey, 2014 Hassan populasi. Available from: https://jurnal.ugm.ac.
of induced abortion in Ethiopia, 2014: changes in the provision of
id/populasi/article/view/44146/24001. Accessed January 3, 2020.
services since 2008. Int Perspect Sex Reprod Health. 2016;42
23. Chola M, Michelo C. Proximate determinants of fertility in Zambia:
(3):111–120. doi:10.1363/42e1816
analysis of the 2007 Zambia demographic and health survey.
14. Bongaarts J. Modeling the fertility impact of the proximate determi-
Int J Popul Res. 2016;2016:1–7. doi:10.1155/2016/5236351
nants: time for a tune-up John Bongaarts. Demogr Res.
24. Tekelab T, Melka AS, Wirtu D. Predictors of modern contraceptive
2015;33:535–560. doi:10.4054/DemRes.2015.33.19
methods use among married women of reproductive age groups in
15. Davis K, JB. Social structure and fertility: an analytic framework.
Western Ethiopia: a community based cross-sectional study. BMC
Econ Dev Cult Chang. 1956;112–135.
Womens Health. 2015;15(1):52. doi:10.1186/s12905-015-0208-z
16. Bongaarts J. A framework for analyzing the proximate determinants
of fertility. Popul Dev Rev. 1978;4(1):105–132. doi:10.2307/1972149

Open Access Journal of Contraception Dovepress

Publish your work in this journal
Open Access Journal of Contraception is an international, peer- studies relating to understanding the biological mechanisms and
reviewed, open access, online journal, publishing original research, practical development of new contraceptive agents. The manuscript
reports, reviews and commentaries on all areas of contraception. In management system is completely online and includes a very quick and
addition to clinical research, demographics and health-related fair peer-review system. Visit http://www.dovepress.com/testimonials.
aspects, the journal welcomes new findings in animal and preclinical php to read real quotes from published authors.

Submit your manuscript here: http://www.dovepress.com/open-access-journal-of-contraception-journal

submit your manuscript | www.dovepress.com

Open Access Journal of Contraception 2020:11 41
DovePress