Time trends in and factors associated with repeat adolescent birth in Uganda: Analysis of six demographic and health surveys.

BACKGROUND: Information on repeat adolescent birth remains scarce in sub-Sahara Africa. We investigated the prevalence and time trends in repeat adolescent birth in Uganda, and associated factors.
METHODS: We analyzed Uganda Demographic and Health Survey data of women age 20-24 years collected on 6 surveys (1988/89-2016) to estimate repeat adolescent birth (first live birth <18 years of age followed by another live birth(s) <20 years). Further, we estimated the wantedness of the second order birth and the prevalence of short birth intervals birth (<13 months) between the first and second such birth. On the 2016 survey, we examined factors associated with repeat adolescent birth using bivariate and multivariate modified Poisson regression.
RESULTS: At the 1988/89 survey, 58.9% of women with first birth <18 years reported a repeat adolescent birth. This percentage increased to 66.8% in 2006 (+7.9 percentage points [pp], p = 0.010) and thereafter declined to 55.6% by 2016 (-11.2 pp, p<0.001), nevertheless, no change occurred between 1988/89 and 2016 (-3.3pp, p = 0.251). Among women with repeat adolescent births, the mean number of live births by exact age 20 years (2.2 births) and prevalence of short birth intervals (3.5% in 1988/89, 5.4% in 2016) (+1.9pp, p = 0.245) did not change. Increasingly more women with repeat adolescent births preferred to have had the second child later, 22.5% in 1995 and 43.1% in 2016 (+20.6pp, p = <0.001). On the 2016 survey, women from poorer households and those of younger age at first birth were significantly more likely to report repeat adolescent birth.
CONCLUSION: Following a first birth <18 years, more than half of the women report a repeat adolescent birth (<20 years), with no decline observed in 30 years. Increasingly more women wanted the second adolescent pregnancy later, highlighting the need to support adolescents with improved family planning services at each contact.

Introduction

Globally, adolescent childbearing remains a major public health concern most especially in the low- and middle-income countries (LMICs). Approximately 1 in 8 of the 140 million births annually occurs to adolescent women with 95% of these occurring in LMIC, and 23% in sub-Sahara Africa [1-4]. Uganda has a high adolescent childbearing rate with estimates at 25% of 15–19 year old having begun childbearing [5]. These levels have remained high in the last 15 years despite a decline in the age specific fertility rate for Uganda among women 15–19 years, 195 births per 1,000 women per year in the 1990 Uganda Demographic and Health Survey (UDHS) to 132 births per 1,000 women per year at the 2016 survey.

Adolescent childbearing lays a foundation for disadvantages in the areas of health, social, and economic outcomes both in the short term and long term [3, 4, 6–9]. Adolescent women and their babies are at higher risk of experiencing poor health outcomes such as obstetric fistula, sepsis, stillbirths, preterm births, birth asphyxia, poor child survival and mental disorders. Socially, adolescent women often face violence from family and community, discrimination and stigma with a high risk of economic disadvantages compounded by premature cessation of schooling and early marriage [10-12]. The younger the adolescent mother, the more vulnerable she is both socio-economically and medically to poor outcomes, including repeat pregnancies [13-15]. Not seldomly, adolescent pregnancies are result of sexual and gender based violence [16-19]. Experiencing another birth before 20 years of age (= repeat adolescent childbirth) may therefore push the adolescent woman and her offspring into worse outcomes than what she experienced following the first birth. Repeat pregnancy in adolescence is more common in settings of high poverty, low educational attainment or its discontinuation, early union or being in a union, none use of long acting reversible contraceptives and previous abortion or non-live birth, among others [20-23].

Information on the extent of repeat adolescent birth, even though it potentially constitutes a large portion of adolescent fertility, is scarce. Second and higher order births among <20 year olds are in some countries a substantial percentage of all such births [24]. Existing studies examined repeat adolescent pregnancies mainly in high income countries with few focusing on repeat adolescent birth, as an end point. These studies, mainly prospective cohorts using health facility samples and systematic reviews, estimated prevalence of repeat pregnancies among adolescents at approximately 17% [21, 22, 25–28]. We identified one study from the Philippines, one from Thailand-Myanmar boarder and three studies from sub-Saharan Africa: Tanzania, South Africa and Uganda [23, 29–32]. The Uganda study, estimated rapid repeat pregnancy (within 12 months) among women 15–22 years at 37% and 74% within 24 months [23]. Whether these repeat pregnancies were wanted then or later, was not determined. Wantedness of pregnancy, though it may not translate into the actual births, provides information on unwanted pregnancies and desired birth intervals [33, 34]. Birth intervals between the first and second birth among women in LMICs, 15–49 years, have lengthened overtime in tandem with fertility decline between 1965 and 2014 [35]. The short birth intervals <24 months declined as the interval between first and second birth increased. Information on initiation of childbearing in adolescence has received some attention in Uganda [23, 36, 37]. Studies have reported a decline in first adolescent childbirth in Uganda [38, 39], but little is known about repeat adolescent birth.

We estimated the levels and time trends of repeat adolescent births (another birth below 20 years) in Uganda using all available six rounds of the UDHS data from 1988/89 to 2016 among women with first birth <18 years of age with a view to inform programming for and delivery of adolescent health care services. We also estimated the percentage of all women age 20–24 years who reported a repeat adolescent birth. The birth intervals between the first and second order adolescent birth and wantedness of the second order birth at that time point were also assessed. Last, using the 2016 UDHS data, we examined factors associated with repeat adolescent birth.

Methods

Data source, population and definitions

We analysed data from all six UDHS rounds (1988/89, 1995, 2000/01, 2006, 2011, and 2016). The DHS are nationally representative cross-sectional surveys that collect information on population, including on maternal and child health. In these surveys, two-stage cluster sampling was conducted with representation of all the geographic regions of a country. All the data was from women’s self-report and was collected by trained data collectors. The interviewer-administered questionnaires used were translated into the local languages and pre-tested prior to data collection. Due to security concerns, 20% of the country was not assessed in 1988/89 and 5% in 2000/01 surveys.

The analysis sample was women aged 20–24 years at the time of each survey. This age category was chosen because of completion of the age at risk of adolescent birth, based on the World Health Organization definition of adolescence (10-19years of age) [40].

Our main outcome was repeat adolescent birth, defined as first live birth <18 years of age followed by another live birth(s) <20 years among 1) women 20–24 years old with first live birth <18 years and 2) among all women 20–24 years. We used live birth as an outcome (rather than pregnancy) as this information is consistently available in the DHS.

Birth intervals were calculated between the first and second order live birth among women with repeat adolescent birth. We categorized the intervals into three groups; <13 months (short birth interval), 13–24 months and above 24 months. All the women had completed the period of observation of 12 months, according to our definitions. Among a subset of women for whom the second birth had occurred within the 5 years preceding the survey (i.e. during the survey recall period during for which this question was asked), we assessed whether the women wanted the pregnancy then, later or wanted no more (= wantedness). Information on wantedness of that birth was available for all surveys except 1988/89.

The 2016 survey data set was chosen for analyzing the factors associated with repeat adolescent birth as this was the most recent survey that would provide the most current information on these factors. The risk factors explored in multivariable analysis were socio-demographic position (region, residence, religion, and household wealth quintile) at the time of the survey, and sexual/reproductive health predictors—age at first sex and age at first birth. Regions were categorized into four based on the categorization at the 2000/01 survey. Central region contained Kampala, Central 1 and Central 2. Northern region included Lango, Acholi and West Nile sub-regions. Eastern region was composed of Teso, Karamoja, Bugisu, Bukedi, and Busoga sub-regions while Western region was composed of Bunyoro, Tooro, Ankole and Kigezi sub-regions. Residence and household wealth quintile were maintained as captured in the survey data set. Religion was re-categorized into 4 categories; Anglican, Catholic, Muslim and Other. Catholic and Muslim religions were not re-categorized. Anglican religion included Anglican and Pentecostal/born again/evangelical. “Other” religion category included all the remaining groups; Seventh day Adventist, orthodox, Baptist, traditional, no religion, other, and Jehovah’s Witness. Due to possibility of reverse causality, the following potential factors were not included in the analysis: education level at time of survey, occupation at time of survey, and marital status at survey. Previous studies have indicated that education level, occupation and marital status are predictors of first birth [41] and therefore, by the time of the second order birth, reverse causality may be seen. The final variables included were chosen based on availability of this information in the DHS data and predictors from previous studies. Contraception use was excluded as there was no information on its use at the point of the repeat adolescent pregnancy. The DHS collects information about contraception use at the point of the survey.

Analysis

Analysis was performed using STATA version 12.0, StataCorp LP, Texas. Sample weights were applied for all analysis to have a sample that is representative of the whole population and reduce sampling bias [42, 43]. The issue of multiple (twins or triplets etc) births was factored in when categorizing into repeat and no repeat adolescent births and calculating the birth intervals by counting each delivery event as one birth, irrespective of multiplicity.

We calculated the absolute percentage point (pp) differences in outcomes between survey points and used the two-sample test of proportions to estimate the p-value of differences. Descriptive statistics for the characteristics using proportions for categorical variables and means with standard deviations for continuous variables were presented. Birth intervals between the first and second birth and wantedness of the second order live birth were presented as proportions.

In determining the factors associated with repeat adolescent birth on the 2016 survey, we calculated the crude and adjusted prevalence risk ratios. The modified Poisson regression was used for this analysis and it was chosen over the logistic regression in this cross sectional study, with a binary outcome, so as to avoid odds ratios over estimating the prevalence ratios in our scenario where the likelihood of the outcome was high, above 10% [44]. All factors were included in the final models irrespective of whether they were significant at crude analysis or not. The total sample of women with first birth <18 years was 1084 and there were no missing values in variables to calculate the outcome. In calculation of the mean age at first sex, 47 women were classified as inconsistent (age at first sex indicated as having occurred after childbirth) and were excluded from this analysis.

Ethics

The School of Medicine Research Ethics Committee (SOMREC) Makerere University and the Uganda National Council for Science and Technology (UNCST) gave ethical approval for the study. Permission to access and use the data sets was sought from the DHS program that collects data after obtaining approvals from the Government of Uganda and informed consent from respondents during the survey.

Results

The prevalence and time trends in repeat adolescent birth

In Table 1, the data in the last row “among women 20–24 years with first birth <18 years” is a sub-set sample of those with a first birth before 18 years (the two rows directly above). Among the sample of all women age 20–24 years, those reporting first birth <18 years reduced from 41.7% (411/985) at the 1988/89 to 28.4% (1084/3822) at 2016 survey. The percentage who reported repeat adolescent birth following a first birth <18 years of age was 24.6% (242/985) at the 1988/89 survey, increased to 26.6% (399/1504) at the 2000/01 and thereafter declined to 15.8% (603/3822) at the 2016 survey (Table 1). Overall, the percentage of the women 20–24 years reporting repeat adolescent birth declined between 1988/89 and 2016 (-8.8 percentage points [pp], p<0.001) (S1 Table).

Table 1

Prevalence and time trends in repeat adolescent birth among Uganda women age 20–24 years (% and 95% CI); all surveys.

Survey	1988/89 UDHS	1995 UDHS	2000/01 UDHS	2006 UDHS	2011 UDHS	2016 UDHS
Among all women 20–24 years (column %, 95% CI)
	N = 985	N = 1555	N = 1504	N = 1710	N = 1629	N = 3822
No birth <18	58.3	60.9	58.1	64.8	67.0	71.6
	(54.5–62.0)	(57.7–64.0)	(54.6–61.4)	(62.1–67.3)	(63.9–69.9)	(69.8–73.4)
1st birth <18, No repeat birth <20	17.1	14.8	15.4	11.7	12.6	12.6
	(14.6–19.9)	(12.8–16.9)	(13.4–17.6)	(10.2–13.3)	(10.8–14.8)	(11.4–13.9)
1st birth <18, Repeat birth <20	24.6	24.3	26.6	23.5	20.4	15.8
	(21.5–28.0)	(21.4–27.5)	(23.7–29.7)	(21.3–26.0)	(17.9–23.2)	(14.4–17.2)
Among women 20–24 years with first birth <18 years (%, 95%CI)
	N = 411	N = 608	N = 631	N = 603	N = 538	N = 1084
Repeat birth <20 years	58.9	62.2	63.3	66.8	61.7	55.6
	(53.5–64.2)	(57.1–67.1)	(58.8–67.6)	(62.8–70.6)	(56.3–66.8)	(52.1–59.1)

Among women 20–24 years with first birth <18 years, 58.9% (242/411) of the women in 1988/89 survey reported a repeat adolescent birth compared to 55.6% (603/1084) at the 2016 (Table 1 and Fig 1). There was no change in repeat adolescent birth in the initial 15 years (+4.4pp, p = 0.154) compared to the later 15 years, 2006–2016 (-11.2pp, p<0.001). Overall, the percentage of women with first birth <18 years reporting repeat adolescent birth in 1988/89 and 2016 were similar- a -3.3pp difference (p = 0.251).

Fig 1

Percent of women age 20–24 years with first childbirth <18 years reporting repeat adolescent birth in Uganda, by year of survey.

Among all the women 20–24 years, the mean number of births by exact age 20 years declined from 0.95 to 0.69 at the 1988/89 and 2016 surveys, respectively (Table 2). The decline in mean and median number of live births started after the 2000/01 survey. Among the women with repeat adolescent birth following first birth <18 years, the mean (2.2) and median (2) number of births by exact age 20 years did not change across the surveys. Further, the percentage of those with repeat adolescent birth reporting 3 or more children by exact age 20 years did not change in the 30 years, 19.8% in 1988/89 and 20.2% in 2016.

Table 2

Mean and median number of live births by age 20 years (<20 years).

Survey	1988/89 UDHS	1995 UDHS	2000/01 UDHS	2006 UDHS	2011 UDHS	2016 UDHS
Among all women 20–24
Mean no. live births (SD)	0.95	0.93	0.97	0.89	0.78	0.69
	(0.9)	(0.9)	(0.9)	(0.9)	(0.9)	(0.8)
Median no. live births (IQR)	1	1	1	1	1	0
	(0–2)	(0–2)	(0–2)	(0–1)	(0–1)	(0–1)
Among women 20–24 with first birth <18 years
Mean no. live births (SD)	1.73	1.79	1.76	1.85	1.77	1.69
	(0.7)	(0.8)	(0.7)	(0.7)	(0.7)	(0.7)
Median no. live births (IQR)	2	2	2	2	2	2
	(1–2)	(1–2)	(1–2)	(1–2)	(1–2)	(1–2)
Among women 20–24 with first birth <18 years and repeat adolescent birth <20 years
Mean no. live births (SD)	2.23	2.27	2.20	2.27	2.24	2.24
	(0.5)	(0.5)	(0.5)	(0.7)	(0.5)	(0.5)
Median no. live births (IQR)	2	2	2	2	2	2
	(2–2)	(1–2)	(2–2)	(2–2)	(2–2)	(2–2)
% of women 20–24 with first birth <18 years and repeat adolescent birth <20 years reporting 3 and more births by exact age 20 years
3 and more children	19.8	22.7	18.7	22.8	22.1	20.2
	(14.3–26.7)	(18.3–27.8)	(14.4–24.0)	(18.5–27.7)	(17.0–28.2)	(16.8–24.0)

*SD–standard deviation, IQR -Interquartile range

Birth interval between first and second birth order

The percentage of women with repeat adolescent birth reporting a birth interval of <13 months was 3.5% (95% CI 1.6–7.7) at the 1988/89 survey with highest proportion being 6.7% at the 2000/01 survey. During the entire period of observation, there was no change in the extent of birth intervals <13 months, with a +1.9pp difference (p = 0.245) in this birth interval between 1988/89 and 2016 surveys. Women reporting a birth interval between 13–24 months remained at just over 40% (Table 3).

Table 3

Birth intervals between first and second live birth among women age 20–24 years with repeat adolescent birth following first birth <18 years, all surveys (column % and 95%CI).

Survey	1988/89	1995	2000/01	2006	2011	2016
Birth interval category	N = 242	N = 378	N = 399	N = 403	N = 327	N = 603
<13 months	3.5	3.4	6.7	4.4	4.0	5.4
	(1.6–7.7)	(1.9–5.8)	(4.0–10.8)	(2.5–7.6)	(2.2–7.3)	(3.8–7.5)
13–24 months	44.3	42.2	48.0	50.0	41.8	42.7
	(37.9–50.8)	(36.2–48.5)	(42.5–53.7)	(44.9–55.1)	(35.0–48.8)	(38.4–47.1)
>24 months	52.2	54.4	45.3	45.7	54.2	51.9
	(45.4–59.0)	(48.1–60.5)	(39.4–51.3)	(40.6–50.8)	(47.0–61.2)	(47.4–56.4)

Wantedness of second order birth

We analysed women with repeat adolescent birth in whom the second live birth occurred in the 5 years preceding the survey and therefore had data on pregnancy wantedness. The percentage of women reporting having wanted that pregnancy to come later increased from 22.5% (95% CI, 15.8–31.1) at the 1995 survey to 43.1% (95% CI, 37.4–49.0) at the 2016 (+20.6pp difference, p <0.001) (Table 4).

Table 4

Wantedness of second adolesecent birth among Uganda women age 20–24 years with first birth <18 years and repeat birth <20, 1995–2016, column % and 95% CI.

Survey	1995	2000/01a	2006	2011	2016
Total Sample with repeat births	N = 378	N = 399	N = 403	N = 332	N = 603
% analyzed for wantedness	48.7%	65.7%	55.8%	65.1%	59.7%
Then	75.7	72.1	58.1	61.8	54.2
	(67.2–82.6)	(64.9–78.3)	(50.5–65.4)	(52.5–70.3)	(48.3–59.9)
Later	22.5	23.1	37.6	38.0	43.1
	(15.8–31.1)	(17.3–30.2)	(30.6–45.1)	(29.5–47.3)	(37.4–49.0)
No more	1.7	4.7	4.3	0.2	2.7
	(0.5–5.4)	(2.1–10.5)	(2.1–8.7)	(0.0–1.6)	(1.3–5.7)

a2006; 0.4% missing

Factors associated with repeat adolescent birth

We analysed factors associated with repeat adolescent birth on the 2016 survey. A total of 1084 women age 20–24 years had a first birth <18 years (Table 5). In the crude associations of the various factors with repeat adolescent birth, rural residents were more likely than urban residents to have had a repeat adolescent birth (crude prevalence ratio [PR] 1.31, 95% CI = 1.08–1.60). Women from Eastern and Western regions were more likely to report the outcome compared to those from central region (PR 1.38, 95% CI = 1.13–1.69 and 1.25, 95% CI = 1.01–1.54, respectively). Each additional year decrease in age at first sex and age at first birth was associated with increased likelihood of reporting a repeat adolescent birth (PR 0.84, 95% CI = 0.81–0.88 and 0.76, 95% CI = 0.74–0.79 respectively). Women from the poorest wealth quintiles were more likely to report the outcome.

Table 5

Factors associated with repeat adolescent birth among Uganda women 20–24 years at survey with first birth <18 years, 2016 UDHS (N = 1084).

	Total	Repeat adolescent childbirth (row %; 95% CI)	Crude PR (95% CI)	P-value (crude)	Adjusted PR (95% CI)	Wald test P-value
Residence
Urban	208	44.4 (36.4–52.7)	1		1
Rural	875	58.3 (54.5–62.0)	1.31 (1.08–1.60)	0.007	1.07 (0.89–1.28)	0.461
Region
Central	278	45.6 (37.7–53.8)	1		1
Eastern	337	63.0 (57.0–68.6)	1.38 (1.13–1.69)	0.002	1.17 (0.97–1.40)	0.096
Northern	216	55.4 (49.3–61.4)	1.21 (0.99–1.50)	0.069	0.97 (0.79–1.19)	0.760
Western	252	56.9 (50.5–63.1)	1.25 (1.01–1.54)	0.040	1.10 (0.91–1.34)	0.328
Religion
Anglican	461	59.9 (54.6–64.9)	1		1
Catholic	416	53.0 (47.5–58.4)	0.88 (0.78–1.01)	0.071	0.97 (0.86–1.10)	0.624
Muslim	172	50.3 (41.5–59.1)	0.84 (0.69–1.02)	0.075	0.97 (0.81–1.15)	0.696
Other	036	56.2 (39.8–71.4)	0.94 (0.69–1.28)	0.687	0.98 (0.74–1.29)	0.865
Household wealth quintile
poorest	278	67.8 (61.7–73.2)	1		1
poorer	250	54.4 (47.7–61.0)	0.80 (0.69–0.93)	0.003	0.84 (0.73–0.96)	0.009
middle	188	61.9 (54.3–69.0)	0.91 (0.79–1.05)	0.216	0.91 (0.78–1.06)	0.241
richer	189	49.0 (40.2–57.9)	0.72 (0.59–0.89)	0.002	0.81 (0.67–0.98)	0.028
richest	179	38.7 (30.1–48.0)	0.57 (0.45–0.73)	<0.001	0.64 (0.48–0.84)	0.001
Age at first sex
Mean age(years (SD)	14.7 (1.42)	14.3 (1.38)	0.84 (0.81–0.88)	<0.001	1.00 (1.00–1.00)	0.968
Age at first birth
Mean age(yrs) (SD)	15.8 (1.25)	15.3 (1.30)	0.76 (0.74–0.79)	<0.001	0.77 (0.74–0.80)	<0.001

In adjusted analysis, two factors were found to be significantly associated with reporting a repeat adolescent birth: household wealth quintile and age at first birth. Women in the richer (adjusted prevalence ratio [aPR] 0.81, 95%CI 0.67–0.98) and richest (aPR 0.64, 95% CI = 0.48–0.84) household wealth quintiles were less likely to report a repeat adolescent birth compared to those in the lower three poorest quintiles. Each additional year increase in age at first birth was associated with a 23% lower likelihood of reporting repeat birth (p <0.001).

Discussion

Prevalence and time trends of repeat adolescent birth

Our study found that approximately half of women 20–24 years reporting first birth <18 years of age had another birth at each of the six UDHS surveys. There was no significant decline in the prevalence of repeat adolescent birth over the 30-year period of observation, including the mean number of births by exact age 20 years among these women. However, in the entire sample of all women age 20–24 years, we found decline in repeat adolescent births as a result of an overall decline in women reporting first birth <18 years. The wantedness of the second order birth declined in this period with—more women report having wanted to delay the second pregnancy.

We defined repeat adolescent birth as a second or higher live birth following a first birth at <18 years among women age 20–24 years, unlike other studies that used different measurements, thereby making direct comparison of results difficult. For example, a study using the 2011 UDHS defined “rapid repeat pregnancy” as any other pregnancy among married or cohabiting women 16–22 years of age with one or two previous pregnancies, irrespective of the previous pregnancy outcome [23]. They reported the prevalence of rapid repeat pregnancy of 37% and 74% within 12 and 24 months, respectively. A study from South Africa evaluated prevalence of repeat pregnancies, as an end point, among 13–19 year old black adolescents who were pregnant, recently delivered or had terminated a pregnancy [29]. The authors found a repeat pregnancy prevalence of 17.6% in the first 24 months of observation. Finally, a study in the Philippines using DHS data defined repeated births as an adolescent 15–19 years with atleast two live births. They did not restrict it to those who had first birth <18 years. The authors reported high repeated births among adolescents 15–18 years with negligible reductions over the 20 years with repeated birth declining from 8.49% in 1993 to 7.80% in 2013 [31]. To compare findings on repeat births in adolescence across countries, it could be helpful to decide on a defined measurement that takes into consideration those initiating birth during the most vulnerable period in adolescence, <18 years of age. Births during younger adolescence, <18 years, carries more disadvantages than in older adolescence [8, 36].

Our results suggest that the current generic programs such as those that aim to increase family planning use, have not addressed prevention/delay of repeat adolescent births among adolescents with first birth <18 years. Over time, there was no change in repeat adolescent birth, highest among women from poor households and those reporting first birth at a younger age- each additional yearly decrease in age at first birth was associated with increased likelihood of reporting a repeat adolescent birth. Previous studies demonstrated that the reaction of families and communities in Uganda following first pregnancy/birth might make girls drop out of school, get married, and therefore set on a pathway into repeat adolescent births [11, 12]. Implementation of the law against marriage <18 years act has had challenges with persistence of high sexual and gender-based violence against adolescent girls, including forced marriage/union [45, 46]. This may be sustaining the persistent repeat adolescent births as girls get sent off into marriage at or before first birth and vice versa. The socio-cultural and political positioning in Uganda appears to be perpetuating adolescent births irrespective of the laws and other programs [47]. Our study results show worst statistics during the 2000/01 and 2006 surveys and this was the period when Uganda rollout the implementation of universal primary education and the defilement act (sex or marriage with a girl <18 years was prohibited under the law)- from 1997. The effects of these policies would perhaps have started to be felt after the 2000 survey-in the younger adolescent age groups that had time to benefit from it. Therefore, based on the sample analysed (women 20–24 years), the cohort that would have benefited from these interventions would have been the one surveyed in 2011. Further, contraceptive uptake among sexually active adolescents in Uganda is low, approximately 12%, despite the commitments to improve family planning programs in the country [48-50]. In comparison to factors associated with first adolescent birth in Uganda, after adjusting for age at first birth, age at first sex was no longer significant, but it was significant in crude analysis [38, 39]. Further, repeat birth in adolescence did not appear to be associated with any of the cultural factors such as area of residence or religion that were significant factors for first adolescent birth [38]. This suggests that perhaps it is not so much about when girls start having sex, but rather about contraceptive use and when they have the first birth. A study using the 2011 UDHS investigated rapid repeat pregnancy among Ugandan women 16–22 years [23] and reported that the factors associated with rapid repeat pregnancy among currently married or cohabiting women at interview were: rural residence, region, and age at first union. Our study did not include age at first union in the analysis. Although the greatest burden of repeat adolescent birth (approximately 84%) was among women from rural residence, residence and region were not significant factors at multivariate analysis.

In other LMIC settings, a study in Philippines using DHS data found that repeated pregnancy was more common among adolescents from poorer communities [31]. This study and another systematic review highlighted the limited information on factors associated with repeat adolescent births in LMICs [21]. Poverty is associated with repeat adolescent pregnancies and births in both LMIC and high-income settings and has a high chance of reverse causality [20, 21]. Poverty perhaps deprives the girl of the power to make decisions over further births, family planning use, access to abortion or by increasing chances of her deciding to complete her family size early due to lack of viable alternatives [20, 21]. Our study further supports that household poverty and young age at first birth appear to be major factors associated with repeat adolescent births.

Birth interval and wantedness of the repeat adolescent birth

The prevalence of very short birth intervals (<13 months) did not change significantly over the 30 years and remained low 5.4% in 2016, probably due to protection from lactational amenorrhea. Breastfeeding is almost universal in Uganda and mean breastfeeding months being 14.1 months [51]. Over 40% of women reported a birth interval of 13–24 months between the first and second order adolescent births. The failure of the birth spacing of 24 months and less to decline among adolescents with first birth <18 years is in contrast to the decline noted globally among women of reproductive age in LMICs [35]. As fertility has declined in LMICs examined, there has been a lengthening of the birth interval between first and second birth among women 15–49 years. This contrasts with our findings of no significant change in short birth intervals, between first and second birth, among this age category of women. This points to a challenge of family planning uptake among adolescents. Other studies have suggested a phenomenon of the adolescent girls making personal decisions to have more children as a means of building families, reinforcing their motherhood identity, and stabilizing their relationships irrespective of their circumstances and available options [52]. However, our study showed that progressively more women wanted to delay the second order birth despite no reductions in prevalence of repeat adolescent birth over the 30 years. These results support the high unmet need for contraception in Uganda, poor access to reproductive health information, weak adolescents’ decision making capacity regarding their reproductive health choices, and short birth intervals, among others [23, 50, 53, 54].

Limitations

Our study used data from cross sectional surveys that collected self-reported information. This may have been affected by response and recall bias. Further, causality cannot be determined using cross sectional data but rather, associations with the inherent challenge of reverse causality. Further, the study did not explore all possible predictors such as partner-related predictors and influence of previous pregnancy outcome. This study, however, is an important starting point to understand the trends and factors associated with repeat adolescent births using nationally representative samples in sub-Saharan Africa. To reduce on potential recall bias, we analysed data among women age 20–24 years that had a shorter recall period for information pertaining events during their adolescence. This age category had completed the period of observation based on the WHO definition of adolescence period. We do acknowledge that other literature suggests that adolescence may stretch to 24 years of age, with no clear cut off age [55]. In this study, we used live births which does not capture all pregnancies- as others end in abortion or stillbirths. Information on induced abortion in Uganda tends to be under reported as it is prohibited other than for prescribed conditions to save the life of a woman [56, 57]. Contraceptive use was not examined and yet previous studies indicated that it is a predictor of repeat adolescent birth [41].

Conclusion

Repeat adolescent birth, among women with first birth <18 years, is high in Uganda (more than 1 in 2 women) with no decline observed in 30 years. The mean number of live births by age 20 years among these women did not decline, remaining at 2.2 births. This contrasts with the finding that the percentage having a first birth <18 years has reduced. Birth intervals have not changed over the 30 years, but the wantedness has–women are much less likely to report that they wanted the second adolescent pregnancy at that time. This together might mean that adolescents would like to be supported in avoiding or delaying the second birth, but that the most vulnerable/young/marginalized adolescent girls are being neglected- the poor and those who had first birth at a young age- with services. This is particularly worrying, because adolescents pregnant with their first child would have been very likely to access antenatal, childbirth, postnatal and child vaccination services, which provide ample opportunities to provide postnatal family planning counselling and services. These missed opportunities may keep young mothers in a spiral of repeat adolescent births and we therefore recommend that, every contact between healthcare workers and pregnant adolescents be utilized as an opportunity to prevent unwanted repeat adolescent birth. We suggest future research explores the circumstances and motivators for the repeat adolescent births. This qualitative research should investigate this among women with and without repeat adolescent birth, their parents and partners. There is need to understand how poverty leads to repeat adolescent births and vice versa. Further, understanding of the implementation of the policies and legislation to protect girls <18 years of age is important as a basis for explaining these high repeat adolescent births. Regarding the birth intervals, analysis needs to examine birth interval for first to second childbirth intervals among women 20–24 years with a first birth at or after 18 years so as to determine whether these birth intervals are typical across women who start childbearing both before 18 years and across women who start child birth at and after 18 years.

Percent point difference in repeat adolescent birth between surveys among Uganda women age 20–24 years, all UDHS surveys.

(PDF)

Click here for additional data file.

3 Feb 2020

PONE-D-19-31884

Time Trends in and factors associated with Repeat Adolescent Birth in Uganda: analysis of six Demographic and Health Surveys

PLOS ONE

Dear Dr Amongin,

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 Mar 19 2020 11:59PM. 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,

Chaisiri Angkurawaranon

Academic Editor

PLOS ONE

Journal Requirements:

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

1. 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.plosone.org/attachments/PLOSOne_formatting_sample_main_body.pdf and http://www.plosone.org/attachments/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[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

**********

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

Reviewer #1: Yes

Reviewer #2: 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: Yes

**********

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: 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: This manuscript quantitatively describes repeat adolescent pregnancies in Uganda over almost 30 years. This is a valuable analysis of a meaningful outcome in a vulnerable population, and certainly deserves publication. Please see comments below. Overall, the data (and other previous data) is looked at from many angles and it is important to clarify and highlight the differences in the different statistics offered – eg what is the denominator? is it a rate or a prevalence? etc. I applaud the authors for tackling this complicated subject and look forward to seeing the final publication.

Major Comments:

1. The authors discuss the lack of a standardized definition of repeat adolescent pregnancy as a challenge. However, they chose not to use the definitions previously used in the literature. Please explain the rationale for your definition and explain why the benefit of doing the analysis in this way was big enough to outweigh the drawback of lack of comparability with other reports. (Or re-run the analysis using methods previously described by other authors.)

2. A major advocacy point that emerges from this data is the need for every contact between healthcare workers and pregnant adolescents to be seen as an opportunity to prevent unwanted repeat adolescent birth. This is powerful and actionable. It is amazing that the overall rate of adolescent birth is falling over time, but the rate of repeat adolescent birth (arguably easier to target) is not.

Minor Comments:

1. There are some minor areas of awkward wording in English. Eg:

a. Line 42 and 48: “More women with repeat adolescent births preferred to have…” consider "Many women with repeat adolescent births..." or "An increasing number of ...”

b. Line 68-69: “yet, the way the sexual intercourse came about may be linked to sexual and gender based violence…” consider: “and many of these pregnancies are result of sexual and gender based violence"

c. Line 318-19 “This qualitative research should understand this” consider: “This qualitative research should investigate this” or “This qualitative research should approach this”

2. Line 57-60 wording is unclear. First half of sentence says no change, the second half says there has been a decline. Consider: “These levels have remained high in the last 15 years despite a decline in the age specific fertility rate for Uganda among this category of women, 195 births per 1,000 women in the 1990 Uganda Demographic and Health Survey (UDHS) to 132 births per 1,000 women at the 2016 survey.” Also, I think these numbers should be “per 1,000 women per year”.

3. Line 79: There are other studies that look at adolescent births and then estimate repeat pregnancies (though not the primary outcome) eg: Parker AL, Parker DM, Zan BN, Min AM, Gilder ME, Ringringulu M, et al. Trends and birth outcomes in adolescent refugees and migrants on the Thailand-Myanmar border, 1986-2016: an observational study. Wellcome Open Research. 2018 May 21;3:62.

4. Line 96: How were surveys conducted? How many languages? In person facilitation? This gives an idea of what into the data collection. Could use a reference if it is described elsewhere.

5. Line 123-125: The decisions on these categories seem a little arbitrary but I don't know the religious dynamics in Uganda intimately. I would expect Baptist might be appropriate to group with Anglican and Penticostal under "Protestant"... Alternately, Anglican and Penticostal are quite different in some parts of the world (but maybe not in Uganda). This may be fine as is but it struck me as a bit odd. The “traditional” religion group seems like a potentially interesting cohort but it may have been very small numbers and therefore hard to look at.

6. Line 127-129: these variables that were excluded due to potential reverse causality seem very interesting.

a. Educational level at a certain point predates the pregnancy – for example, “no education” or “some primary school” or “completed primary school” could be interesting variables that would be expected to pre-date the first pregnancy at 14-<18 yo.

b. Occupation of the woman in her early 20’s could be an interesting “outcome” variable, understanding that the association is complex and may be two-way.

7. Methods: Line 159 and onward – please provide the actual numbers in parentheses with the percentages. This helps the reader track the denominators of the different percentages.

8. Table 1: suggest changing “1st birth <18, Repeat” to “1st birth <18, Repeat birth <20” for clarity

9. Table 2: I'm not sure this table adds anything. It is quite busy and a bit confusing. The bottom line is that there is a pretty consistent downward trend in repeat adolescent birth over the time period, but it is not as consistent if you look at just the percentage of women with adolescent birth who go on to have a second adolescent birth - this is an important potential area for intervention. I think this could be more compellingly demonstrated using a graph.

10. Table 3: Again, this is complicated to present. I'm sure number of births was not normally distributed, so numerically presenting mean and SD is not particularly meaningful. Again, graphing means might give a nice visual.

11. Table 5: bottom rows - I think this is not necessary to include... or could go in the supplement. It is obvious that women at the older end of the age range are less likely to have had their children as adolescents within the previous 5 years.

12. Table 6: (and line 267-269) Urban/Rural residence and region both look highly significant in the univariate analysis. Is there a problem with collinearity between these variables that leads to loss of significance in the multivariable analysis? What happens if you include only one (rural/urban or region) in the analysis – do you retain the effect?

13. Discussion: Most of the statistics look the worst during the 2000 and 2006 surveys. Are there any policy or cultural pressures during this time (or before or after) that could explain these changes? It would be interesting to have some context here.

14. Line 257: you talk about the fact that some girls desire this second adolescent pregnancy, but this data shows a large proportion do want to delay. This is an important advocacy point that should be highlighted (and gets a little lost).

15. Line 280-282: Good point. It would be nice to mention breastfeeding rates in Uganda.

16. Line 302-303: abortion (and to a lesser extent the more rare outcome of stillbirth) is an interesting limitation. A comment about whether or not termination of pregnancy (as opposed to spontaneous abortion) is available/legal in Uganda might give some helpful background.

17. Line 319-20: “There is need to understand how poverty leads to repeat adolescent births…” Suggest adding “and vise versa.”

18. Line 322-323: “Early child birth is a trap for many births, and this hasn’t change over time- this needs to be studied more.” This final sentence is a bit abrasive and negative. Other than further study, it doesn’t suggest any action to redeem this problematic situation. I would suggest: "Pregnancy and child birth increase contact with the health care system, and each contact should be seen as an opportunity for education and services to prevent subsequent unwanted pregnancies."

Reviewer #2: 1. Review summary

• This reviewed paper aims to understand trends and factors of repeated births among adolescents in Uganda, where adolescent childbirths are prevalent. The authors’ paper provides a solid contribution to a scant evidence base on repeated adolescent pregnancies in sub-Saharan Africa. As the authors acknowledge, while much has been studied about early childbearing among adolescents, less is known about adolescent childbearing frequency and timing following an initial childbirth.

• For the most part, the arguments framed in the introduction are clear and support the rationale for the methods and analyses undertaken. The claims made in the conclusion are also mostly supported by the data presented in the results. There are however, a number of revisions that could be made to the analysis and framing of the paper, further described below.

2. Overall suggestions for improvements

• There are important revisions that could significantly improve the quality and rigor of the analysis. Broadly these include:

o Focusing the analysis and results on descriptive trends and risk factors of repeated births and pregnancy spacing only (and removing data analysis on pregnancy wantedness). The inclusion of all three outcomes can be confusing and distracting to the reader, and takes away from the overall message and research question focused on childbirth frequency.

o If all three trends analyses are kept, then a clearer connection should be made in the introduction justifying why all three outcomes are being used in the analysis (particular in using ‘pregnancy wantedness’ to show that women often don’t want their second pregnancy).

o Additional and/or clearer information should be provided in the introduction, analysis and discussion about typical birth spacing practices within adolescent and adult populations, and how these spacing practices might affect childbirth repetition among adolescents. For example, it is unclear from this analysis whether repeat pregnancy is in fact a problem in this adolescent population, or whether repeat pregnancy among adolescents is simply reflective of typical birth spacing practices (if two year birth spacing intervals are typical, then it would make sense that a large proportion of women who have a first child birth prior to 17 years will inevitably have another birth prior to 20 years). It appears that the authors attempt to address this by providing additional analyses on birth spacing intervals in the results, but the connection and implications is still unclear for the reader.

o Since this is a trends analysis, the authors should consider including additional risk factor models examining how the different risk factors change (or do not) across different DHS survey rounds (see more specific suggestion below).

3. Suggestions for improvements by section

3.1. Introduction:

• The introduction could be strengthened by including additional information about current birth interval spacing in youth versus those of adult populations, in addition to current literature around pregnancy desires/wantedness (if you keep all three types of analyses in your paper)

• The consequences of repeated births are very well explained in the introduction, but no background information is provided on the risk factors of repeated pregnancies. This information should be included in the introduction to justify your variables selection in the risk factor analysis later on.

3.2. Methods and analysis:

• The rationale for using the DHS 2016 survey round for the risk factor analysis should be explicitly stated (rather than other rounds)

• The risk factor analysis is interesting – if the data are available, you could expand on this analysis by using repeated cross-sectional surveys to assess changing risk factors over time to see how or if predictors have changed over time

• Either in the introduction or methods, you should provide a justification for your choice of variables in the final risk factor model

• Please elaborate what your concerns are about reverse causality (1-2 sentences suffice).

• You may want to explain why other variables widely available in DHS were not included (for example contraceptive use, which you mention frequently in the discussion section). You could also include this in your ‘Limitations’ section.

• The use of the prevalence ratio makes sense and is appropriate. However, the rationale behind using poisson regression is unclear (perhaps 1-2 sentences explaining why poisson, rather than logistic regression is necessary could be helpful to the reader).

3.3. Results

• The contents of Table 1 should be explained prior to the sharing of table results in the text. The text and tables can be confusing to read. For example, in Table 1, it should be clarified that the data in the last row “among women 20-24 years with first birth <18 years” is a sub-set sample of those with a first birth before 18 years (the two rows directly above). This took me a bit to understand and sort out without any explanatory text.

• Along with the above point, a clearer explanation of the contents in Table 2 will help clarify any confusions to the reader. For example, it was unclear what the intervals were for ‘5 years’ versus ‘first 15 years’ versus ‘last 15 years’ versus ’30 years’ in Table 2. Alternatively, you could explore presenting the data in Table 2 in a graph form.

• You could potentially discard the information in Table 3 and in the text related to the mean/median number of births – this does not seem to provide much more valuable information and is not mentioned again in your paper later on.

• The birth interval table (Table 4) could be strengthened by including data for first to second childbirth intervals among women 20-24 years with a first birth at or after 18 years. This could help the reader understand whether these birth intervals are typical across women who start childbearing both before 18 years and across women who start child birth at and after 18 years (However, you may have to expand the sample age range for this additional analysis to include women who are older than 20-24 years).

3.4. Discussion:

• If you keep the ‘ pregnancy wantedness’ analyses, please better elaborate in your discussion how these data help validate your findings on repeated pregnancies among adolescents.

• On lines 238-239, you suggest that a defined measurement might be needed. What definition might you recommend based on the measures you used and analysis you conducted?

3.5. Smaller edits:

• In general, the language is clear. However, there are a number of grammatical errors which should be reviewed and corrected (for example ‘ration’ instead of ‘ratio’)

• The use of ‘wantedness’ can be confusing, perhaps ‘pregnancy desire’ could be used instead (if appropriate)

• Lines 68-72 – the logic of these sentences is not clear

• Line 58 – Recommend including the data for the adolescent fertility rate here (in addition to the childbirth rates you have now)

• Line 105 – I believe you mean “among women 20-24 years old with a first live birth”

• Line 132-133 – Please explain what type of bias this might induce

**********

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: Mary Ellen Gilder

Reviewer #2: Yes: Esther J. Spindler

[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.

10 Mar 2020

09th/03/2020

To

Dr. Chaisiri Angkurawaranon

Academic Editor

PLOS ONE

Dear Dr. Angkurawaranon,

Re; Response to reviewers’ comments and resubmission of revised manuscript ID PONE-D-19-31884

Time Trends in and factors associated with Repeat Adolescent Birth in Uganda: analysis of six Demographic and Health Surveys

Thank you for reviewing and providing feedback on this manuscript. Please receive the revised copy with specific responses and changes summarized in the table below.

Reviewers comment Response to comment Line number

Journal Requirements:

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

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

Thank you for this guidance. We have updated the entire manuscript and we believe it meets the journal requirements NA

5. Review Comments to the Author

Reviewer One:

This manuscript quantitatively describes repeat adolescent pregnancies in Uganda over almost 30 years. This is a valuable analysis of a meaningful outcome in a vulnerable population, and certainly deserves publication. Please see comments below. Overall, the data (and other previous data) is looked at from many angles and it is important to clarify and highlight the differences in the different statistics offered – eg what is the denominator? is it a rate or a prevalence? etc. I applaud the authors for tackling this complicated subject and look forward to seeing the final publication.

Thank you so much for this feedback and comments. We have proceeded to clarify on the different statistics offered, in the details below.

Statistics offered for trends are prevalence and the different denominators are clarified in the methods section for each sub-analysis. Pages 5-8

Major Comments: ONE

1. The authors discuss the lack of a standardized definition of repeat adolescent pregnancy as a challenge. However, they chose not to use the definitions previously used in the literature. Please explain the rationale for your definition and explain why the benefit of doing the analysis in this way was big enough to outweigh the drawback of lack of comparability with other reports. (Or re-run the analysis using methods previously described by other authors.) Thank you so much for this observation. Previous studies examined getting/reporting a repeat adolescent pregnancy as an end point and did not examine pregnancy outcome. This is a challenge as some pregnancies get aborted or end in stillbirths. An outcome of a repeat live birth is of major importance as it leaves the adolescent mother with another offspring to care for.

Younger adolescents, <18 years, face more disadvantages than older adolescents. This group is more vulnerable and a second or higher birth among this category of adolescents, while still in adolescence, may compound the disadvantages for the woman and offspring.

We based repeat birth on live births (rather than pregnancies) because that is what is available from the DHS over the many surveys. Lines 118-120, page 6

Lines 283-287, page 15

Major Comments: TWO

2. A major advocacy point that emerges from this data is the need for every contact between healthcare workers and pregnant adolescents to be seen as an opportunity to prevent unwanted repeat adolescent birth. This is powerful and actionable. It is amazing that the overall rate of adolescent birth is falling over time, but the rate of repeat adolescent birth (arguably easier to target) is not. Thank you for this very important advocacy point. We have emphasized it in the abstract and conclusion section. Line 48-50, page 3

Line 377-380,

Pages 19 & 20

Minor Comments:

1. There are some minor areas of awkward wording in English. Eg:

a. Line 42 and 48: “More women with repeat adolescent births preferred to have…” consider "Many women with repeat adolescent births..." or "An increasing number of ...”

Thank you for noting this. We have made the correction. Line 42, page 2

Line 48, page 3

b. Line 68-69: “yet, the way the sexual intercourse came about may be linked to sexual and gender based violence…” consider: “and many of these pregnancies are result of sexual and gender based violence"

Thank you. This has been corrected. Line 68-69, page 4

c. Line 318-19 “This qualitative research should understand this” consider: “This qualitative research should investigate this” or “This qualitative research should approach this” This adjustment has been made. Line 381 & 382, page 20

2. Line 57-60 wording is unclear. First half of sentence says no change, the second half says there has been a decline. Consider: “These levels have remained high in the last 15 years despite a decline in the age specific fertility rate for Uganda among this category of women, 195 births per 1,000 women in the 1990 Uganda Demographic and Health Survey (UDHS) to 132 births per 1,000 women at the 2016 survey.” Also, I think these numbers should be “per 1,000 women per year”.

Thank you so much. This has been corrected.

Line 57-60,

page 3

3. Line 79: There are other studies that look at adolescent births and then estimate repeat pregnancies (though not the primary outcome) eg: Parker AL, Parker DM, Zan BN, Min AM, Gilder ME, Ringringulu M, et al. Trends and birth outcomes in adolescent refugees and migrants on the Thailand-Myanmar border, 1986-2016: an observational study. Wellcome Open Research. 2018 May 21;3:62.

Thank you so much for this literature. We have taken this into consideration and cited it. Lines 83-85, page 4

4. Line 96: How were surveys conducted? How many languages? In person facilitation? This gives an idea of what into the data collection. Could use a reference if it is described elsewhere.

Thank you for noting this. This information has been added Line 108-111,

Page 5 & 6

5. Line 123-125: The decisions on these categories seem a little arbitrary but I don't know the religious dynamics in Uganda intimately. I would expect Baptist might be appropriate to group with Anglican and Penticostal under "Protestant"... Alternately, Anglican and Penticostal are quite different in some parts of the world (but maybe not in Uganda). This may be fine as is but it struck me as a bit odd. The “traditional” religion group seems like a potentially interesting cohort but it may have been very small numbers and therefore hard to look at.

Thank you for this question. We did this based on the Uganda dynamics for which the minority group were placed under “other”. The Anglican and protestant are similar and are second to Catholics in number. Third is Muslims. We discussed this decision among the 4 Ugandan authors on this study, two of whom are social scientists and provided detailed feedback on the appropriateness of this categorization. NA

6. Line 127-129: these variables that were excluded due to potential reverse causality seem very interesting.

a. Educational level at a certain point predates the pregnancy – for example, “no education” or “some primary school” or “completed primary school” could be interesting variables that would be expected to pre-date the first pregnancy at 14-<18 yo.

b. Occupation of the woman in her early 20’s could be an interesting “outcome” variable, understanding that the association is complex and may be two-way.

This is a very important observation. Indeed we discussed this issue within the co-author team, and feel we made the best decision given the limitations of cross-sectional survey data. Education level and occupation can be both predictors and outcomes of first birth; we cannot examine such complex relationships from these datasets, unfortunately. Given that we wanted to produce rigorous results and the fact that at the point of survey-when the women were 20-24 years, we thought it would not be rigorous to include them as predictors of repeat adolescent birth (or in fact, confounders for the other variables in the multivariable model), due to potential reverse causality. We agree with the reviewer that it would be great to look at them at the point of second birth- prospectively, in future studies. NA

7. Methods: Line 159 and onward – please provide the actual numbers in parentheses with the percentages. This helps the reader track the denominators of the different percentages.

Thank you so much. We have included the actual numbers. Lines 182-194, page 9

8. Table 1: suggest changing “1st birth <18, Repeat” to “1st birth <18, Repeat birth <20” for clarity We have made this clarification. Table 1, page 10

Line 195

9. Table 2: I'm not sure this table adds anything. It is quite busy and a bit confusing. The bottom line is that there is a pretty consistent downward trend in repeat adolescent birth over the time period, but it is not as consistent if you look at just the percentage of women with adolescent birth who go on to have a second adolescent birth - this is an important potential area for intervention. I think this could be more compellingly demonstrated using a graph.

Thank you so much for this guidance. This table summarizes the percent point differences between surveys, the initial 15 years, last 15 years and overall difference. We feel it would be essential for the reader to easily access this information and it has been moved to a supplementary material- S1 Table, attached.

Further, we have provided a graph for repeat adolescent birth following first birth <18 years and maintained details in table 1, last row. Lines 187 & 188,

Page 9

S1 Table- attachment

10. Table 3: Again, this is complicated to present. I'm sure number of births was not normally distributed, so numerically presenting mean and SD is not particularly meaningful. Again, graphing means might give a nice visual.

We thank the reviewer for this important point. While we agree that a visual might be more aesthetically pleasing, we believe that some of our readers are interested in the precise estimates and confidence intervals (SD, IQR, etc), which is why we would strongly prefer to keep this table as is.

In terms on the distribution, we present number of births before age 20 among women who are 20-24 years old, so there is not a range of values of parity, like there might be among a population of women in reproductive age. We share the reviewer’s concern about how this distribution is presented, and we therefore show several statistics, including mean, median (among various populations by adolescent fertility pattern), and % with 3+ births. We hope that this captures the breadth of descriptives on this population. NA

11. Table 5: bottom rows - I think this is not necessary to include... or could go in the supplement. It is obvious that women at the older end of the age range are less likely to have had their children as adolescents within the previous 5 years. This guidance is noted. We have deleted those rows and this table is now table 4. Line 228- Table 4,

page 12

12. Table 6: (and line 267-269) Urban/Rural residence and region both look highly significant in the univariate analysis. Is there a problem with collinearity between these variables that leads to loss of significance in the multivariable analysis? What happens if you include only one (rural/urban or region) in the analysis – do you retain the effect? Thank you for noting this. There is no collinearity as each of the five regions has both urban and rural clusters. The variable responsible for the loss of effect is “age at first birth”. NA

13. Discussion: Most of the statistics look the worst during the 2000 and 2006 surveys. Are there any policy or cultural pressures during this time (or before or after) that could explain these changes? It would be interesting to have some context here.

During this period is when Uganda rollout the implementation of universal primary education and the defilement act (sex or marriage with a girl <18 years was prohibited under the law)- from 1997. The effects of these policies would perhaps have started to be felt after the 2000 survey-in the younger adolescent age groups that had time to benefit from it. Therefore, based on the sample analyzed (women 20-24 years), the cohort that would have benefited from these interventions would have been the one surveyed in 2011.

We have added this context in the discussion section. Lines 300-306, page 16

14. Line 257: you talk about the fact that some girls desire this second adolescent pregnancy, but this data shows a large proportion do want to delay. This is an important advocacy point that should be highlighted (and gets a little lost). Thank you for bringing this important point to our attention. We have highlighted this advocacy point in the discussion section. Line 341-346, page 18

15. Line 280-282: Good point. It would be nice to mention breastfeeding rates in Uganda.

Thank you. We have added information on breastfeeding. Line 333 & 334, page 18

16. Line 302-303: abortion (and to a lesser extent the more rare outcome of stillbirth) is an interesting limitation. A comment about whether or not termination of pregnancy (as opposed to spontaneous abortion) is available/legal in Uganda might give some helpful background.

Thank you. We have added this information. Line 361-364, page 19

17. Line 319-20: “There is need to understand how poverty leads to repeat adolescent births…” Suggest adding “and vise versa.”

We have done this. Thank you Line 383, page 20

18. Line 322-323: “Early child birth is a trap for many births, and this hasn’t change over time- this needs to be studied more.” This final sentence is a bit abrasive and negative. Other than further study, it doesn’t suggest any action to redeem this problematic situation. I would suggest: "Pregnancy and child birth increase contact with the health care system, and each contact should be seen as an opportunity for education and services to prevent subsequent unwanted pregnancies."

Thank you so much for this guidance. We have corrected this. Lines 377-380, page 19 & 20

Reviewer #2:

1. Review summary

• This reviewed paper aims to understand trends and factors of repeated births among adolescents in Uganda, where adolescent childbirths are prevalent. The authors’ paper provides a solid contribution to a scant evidence base on repeated adolescent pregnancies in sub-Saharan Africa. As the authors acknowledge, while much has been studied about early childbearing among adolescents, less is known about adolescent childbearing frequency and timing following an initial childbirth.

• For the most part, the arguments framed in the introduction are clear and support the rationale for the methods and analyses undertaken. The claims made in the conclusion are also mostly supported by the data presented in the results. There are however, a number of revisions that could be made to the analysis and framing of the paper, further described below. Thank you very much for this feedback and all the suggestions for improvement. We have addressed the revisions as below. NA

2. Overall suggestions for improvements

• There are important revisions that could significantly improve the quality and rigor of the analysis. Broadly these include:

o Focusing the analysis and results on descriptive trends and risk factors of repeated births and pregnancy spacing only (and removing data analysis on pregnancy wantedness). The inclusion of all three outcomes can be confusing and distracting to the reader, and takes away from the overall message and research question focused on childbirth frequency.

Thank you so much for this important observation and guidance. We analyzed wanteness as part and parcel of birth spacing because it can explain the spacing information obtained. Further, it also speaks to the trends observed. We therefore feel, this information adds to advocacy and therefore programming for this group of women.

However, as highlighted by the reviewer, our most important findings relate to the prevalence of repeat births over time and the predictors, which are the crucial points in our Discussion section. NA

o If all three trends analyses are kept, then a clearer connection should be made in the introduction justifying why all three outcomes are being used in the analysis (particular in using ‘pregnancy wantedness’ to show that women often don’t want their second pregnancy).

Thank you so much for this. We analyzed wantedness as a descriptor. We have provided more clarification in the introduction on its relatedness to the trends and birth intervals Lines 86-89, page 4 & 5.

o Additional and/or clearer information should be provided in the introduction, analysis and discussion about typical birth spacing practices within adolescent and adult populations, and how these spacing practices might affect childbirth repetition among adolescents. For example, it is unclear from this analysis whether repeat pregnancy is in fact a problem in this adolescent population, or whether repeat pregnancy among adolescents is simply reflective of typical birth spacing practices (if two year birth spacing intervals are typical, then it would make sense that a large proportion of women who have a first child birth prior to 17 years will inevitably have another birth prior to 20 years). It appears that the authors attempt to address this by providing additional analyses on birth spacing intervals in the results, but the connection and implications is still unclear for the reader. Thank you for this helpful feedback. We have provided additional information on birth spacing among adolescents compared to adult women - in the Introduction and Discussion sections. Lines 89-92, page 4 & 5.

Lines 337-340, page 18

o Since this is a trends analysis, the authors should consider including additional risk factor models examining how the different risk factors change (or do not) across different DHS survey rounds (see more specific suggestion below).

This is a very important recommendation and we are considering this for another paper. We have discussed this in the co-author team and feel that risk factor analyses across surveys do not fit within the objective and scope of this paper. NA

3. Suggestions for improvements by section

3.1. Introduction:

• The introduction could be strengthened by including additional information about current birth interval spacing in youth versus those of adult populations, in addition to current literature around pregnancy desires/wantedness (if you keep all three types of analyses in your paper) Thank you. We have done this Lines 86-92, page 4 and 5.

• The consequences of repeated births are very well explained in the introduction, but no background information is provided on the risk factors of repeated pregnancies. This information should be included in the introduction to justify your variables selection in the risk factor analysis later on.

We have added this information. Line 72-75, page 4

3.2. Methods and analysis:

• The rationale for using the DHS 2016 survey round for the risk factor analysis should be explicitly stated (rather than other rounds)

We chose the most recent survey, 2016, for this analysis to tease out the most current risk factors for intervention purposes and to inform policy-makers in Uganda (who have indeed requested this analysis). We have added this rationale. Line 128-130, page 6

• The risk factor analysis is interesting – if the data are available, you could expand on this analysis by using repeated cross-sectional surveys to assess changing risk factors over time to see how or if predictors have changed over time

Yes, these data are available. We do request that this be analyzed for another paper. NA

• Either in the introduction or methods, you should provide a justification for your choice of variables in the final risk factor model

We chose variables based on previous literature and on what was available in the DHS data. We have added this justification. Line 146-147, page 7

• Please elaborate what your concerns are about reverse causality (1-2 sentences suffice).

Thank you. This has been done. Lines 144-146, page 7

• You may want to explain why other variables widely available in DHS were not included (for example contraceptive use, which you mention frequently in the discussion section). You could also include this in your ‘Limitations’ section.

We have added this explanation in the methods section and limitations section. Briefly, the DHS measure contraceptive use at the time of survey, so this happens after the first and repeat adolescent childbirth and can therefore not be a risk factor. Lines 147-149, page 7

Lines 363-364, page 19

• The use of the prevalence ratio makes sense and is appropriate. However, the rationale behind using poisson regression is unclear (perhaps 1-2 sentences explaining why poisson, rather than logistic regression is necessary could be helpful to the reader).

Thank you so much for this observation. In scenarios where the likelihood of the outcome is high, above 10%, the modified Poisson regression if preferred to the logistic regression to avoid odds ratios overestimating the prevalence ratios. We did provide this explanation and a reference. Line 162-165, page 8

3.3. Results

• The contents of Table 1 should be explained prior to the sharing of table results in the text. The text and tables can be confusing to read. For example, in Table 1, it should be clarified that the data in the last row “among women 20-24 years with first birth <18 years” is a sub-set sample of those with a first birth before 18 years (the two rows directly above). This took me a bit to understand and sort out without any explanatory text.

Thank you. This explanation has been provided in text and we have also provided additional detail in the row headings. Lines 180-181, page 9

Table 1

• Along with the above point, a clearer explanation of the contents in Table 2 will help clarify any confusions to the reader. For example, it was unclear what the intervals were for ‘5 years’ versus ‘first 15 years’ versus ‘last 15 years’ versus ’30 years’ in Table 2. Alternatively, you could explore presenting the data in Table 2 in a graph form. Thank you so much for this guidance, we have opted to have this as a supplementary table. This will enable readers easily access the percent point differences between surveys NA

• You could potentially discard the information in Table 3 and in the text related to the mean/median number of births – this does not seem to provide much more valuable information and is not mentioned again in your paper later on.

Thank you for this suggestion. We feel this provides valuable information on the distribution of the numbers of births, by adolescent fertility pattern. We would be willing to consider placing this table into Supplementary material given editorial guidance. At the moment, we added a few sentences to the text referring to this finding in more detail. NA

• The birth interval table (Table 4) could be strengthened by including data for first to second childbirth intervals among women 20-24 years with a first birth at or after 18 years. This could help the reader understand whether these birth intervals are typical across women who start childbearing both before 18 years and across women who start child birth at and after 18 years (However, you may have to expand the sample age range for this additional analysis to include women who are older than 20-24 years).

Thank you for this very important observation. We however wanted to restrict our analysis to the trend in birth intervals among this category of women in order to assess if there have been any changes within this category. Expanding it as suggested is a very important recommendation and we have included this in the suggestions for future analysis as it is outside the scope of this paper. Lines 385-389, page 20

3.4. Discussion:

• If you keep the ‘ pregnancy wantedness’ analyses, please better elaborate in your discussion how these data help validate your findings on repeated pregnancies among adolescents. Thank you. We have elaborated this in the discussion section. Line 341-346, page 18

• On lines 238-239, you suggest that a defined measurement might be needed. What definition might you recommend based on the measures you used and analysis you conducted?

We recommend a definition that looks at those initiating birth <18 years, the most vulnerable group in early adolescence. We have included this information. Line 283-287, page 15

3.5. Smaller edits:

• In general, the language is clear. However, there are a number of grammatical errors which should be reviewed and corrected (for example ‘ration’ instead of ‘ratio’) Thank you so much for this observation. We have corrected the grammatical errors. Lines 244 and 253, page 13

• The use of ‘wantedness’ can be confusing, perhaps ‘pregnancy desire’ could be used instead (if appropriate)

This word has been used by other publications using DHS data and the DHS also looks at wanting then, later or not all. We opted to retain it as wantedness.

Pregnancy desire might be erroneously interpreted as measured while the woman was pregnant; while on the DHS this is a retrospective question asking women to recall pregnancy wantedness up to 5 years after they had a live birth. NA

• Lines 68-72 – the logic of these sentences is not clear

These sentences position our age category, having births <18 years, and the fact that they are the most vulnerable to sexual abuse. NA

• Line 58 – Recommend including the data for the adolescent fertility rate here (in addition to the childbirth rates you have now) These were the adolescent fertility rates (15-19 years). We have corrected this. Line 58, page 3

• Line 105 – I believe you mean “among women 20-24 years old with a first live birth”

Yes, we have corrected this Line 117, page 6

• Line 132-133 – Please explain what type of bias this might induce We have done so Line 153, page 7

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

26 Mar 2020

Time Trends in and factors associated with Repeat Adolescent Birth in Uganda: analysis of six Demographic and Health Surveys

PONE-D-19-31884R1

Dear Dr. Amongin,

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,

Chaisiri Angkurawaranon

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. 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 #2: (No Response)

**********

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

Reviewer #2: (No Response)

**********

4. 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 #2: (No Response)

**********

5. 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 #2: (No Response)

**********

6. 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 #2: (No Response)

**********

7. 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 #2: Yes: Esther J Spindler

30 Mar 2020

PONE-D-19-31884R1

Time Trends in and factors associated with Repeat Adolescent Birth in Uganda: analysis of six Demographic and Health Surveys

Dear Dr. Amongin:

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. Chaisiri Angkurawaranon

Academic Editor

PLOS ONE