Association between Maternal High-Risk Fertility Behavior and Childhood Morbidity in Bangladesh: A Nationally Representative Cross-Sectional Survey

Abstract.

In this cross-sectional study, we evaluated data from the 2014 Bangladesh Demographic and Health Survey (BDHS), which consisted of 7,707 married women aged 15–49 years who lived with at least one child younger than 5 years. This study’s primary aim was to examine the relationship between maternal high-risk fertility behavior and child morbidity. To define high-risk fertility behaviors, we considered three variables: maternal age at the time of delivery, birth order, and birth interval. The main outcome measures were mortality-related disease in the past 2 weeks (acute respiratory infection [ARI], diarrhea, and fever) and low birth weight (LBW). We used modified Poisson regression with generalized estimating equations to assess the relationships between the variables of interest. Results indicate that a substantial portion of women (34%) exhibited high-risk fertility patterns; 28.7% engaged in a single high-risk behavior and 5.4% engaged in multiple high-risk behaviors. After adjusting for relevant covariates, high-risk fertility behaviors were significantly associated with an increased likelihood of ARI (adjusted relative risk [ARR]: 1.22, 95% CI: 1.05–1.50), diarrhea (ARR: 1.18, 95% CI: 1.03–1.35), fever (ARR: 1.29, 95% CI: 1.11–1.58), and LBW (ARR: 1.27, 95% CI: 1.10–1.52). In addition, engaging in multiple high-risk fertility behaviors appeared to have far-reaching consequences on the outcomes measured. Maternal high-risk fertility behaviors are important predictors of morbidity in children younger than 5 years. Preventing high-risk fertility behavior may reduce childhood morbidity and mortality in Bangladesh.

INTRODUCTION

Despite substantial improvements to child health outcomes over the last few decades, around 15,000 children younger than 5 years still die every day (5.6 million a year), and most of those deaths are caused by readily preventable diseases.¹ Unless this situation improves, more than 60 countries will fail to achieve United Nations Sustainable Development Goal 3 (SDG3), which aims to end preventable newborn deaths by 2030. Infant and child mortality rates remain especially high in sub-Saharan Africa and South Asia; 80% of all under-5 deaths occur in these regions.¹ More specifically, nearly 3,000 newborns still die each day in South Asia.² At 45 deaths per 1,000 live births, South Asia’s under-5 mortality is second only to the African region.¹

Bangladesh has made significant improvements in child health and has achieved its Millennium Development Goal (MDG) 4 (reducing child mortality).³ At 34.4 deaths per 1,000 live births, Bangladesh’s 2017 under-5 mortality rate was much lower than that of other South Asian countries, including India (39 deaths per 1,000 live births) and Pakistan (75 deaths per 1,000 live births).¹ Nonetheless, the incidence of childhood morbidity in Bangladesh is still alarming.⁴ Indeed, Bangladesh has yet to achieve SDG 3. Despite effective interventions aimed at reducing childhood mortality, infectious diseases such as pneumonia and diarrhea remain major causes of death in children.^1,5 More specifically, of Bangladesh’s 119,000 under-5 deaths, pneumonia and diarrhea account for around 15% and 6%, respectively.⁶ Identifying the risk factors for these illnesses will be critically important for reducing childhood morbidity as well as mortality both in Bangladesh and globally.

In developing countries like Bangladesh, a multitude of socioeconomic and household-level factors are associated with childhood morbidity. For example, younger, female, and rural-dwelling children are more likely to experience morbidity.^7–10 In addition, maternal and paternal characteristics such as education level, occupation, and media access are associated with poor child health outcomes. Similarly, household characteristics such as poverty, unsafe drinking water, and poor sanitation predict increased childhood morbidity.^9,11–16

Maternal high-risk fertility behavior is a biodemographic factor that negatively affects child health. High-risk fertility behaviors generally include too-early or too-late childbearing, a higher number of live births and short birth intervals. A number of studies have already documented the link between various components of high-risk fertility behaviors and negative child health. For example, motherhood at either a very early (< 18 years) or advanced age (> 34 years) is associated with an increased likelihood of under-5 morbidity and mortality both in Bangladesh and other countries.^17–27 Women who begin childbearing at an early age often have a greater number of children,²⁸ and this, in turn, is linked to adverse maternal, infant, and child health outcomes.²⁹ Short birth intervals (< 24 months)^30,31 and higher birth order³² are also associated with infant and child mortality.

To date, all studies on the relationship between high-risk fertility behavior and child health have analyzed each fertility behavior in isolation. That is, no studies examine the influence that a combination of high-risk fertility behaviors has on childhood morbidity. Thus, we aimed to understand the impact that high-risk fertility behaviors have on child health, both separately and in combination. More specifically, using a nationally representative sample, this study assesses whether a relationship exists between maternal high-risk fertility behaviors and acute respiratory infection (ARI), diarrhea, fever, or low birth weight (LBW) in children born within the past 5 years. Identifying such relationships will be crucial for developing effective prevention programs in Bangladesh.

MATERIALS AND METHODS

Data.

This cross-sectional study analyzed data from the 2014 Bangladesh Demographic and Health Survey (BDHS), a nationally representative sample of child-bearing aged women. The BDHS was collected using a two-stage stratified sampling procedure. First, 600 primary sampling units (PSUs) were selected with probability proportional to PSU size (207 in urban and 393 in rural areas). Second, using a systematic sampling procedure, an average of 30 households were selected per PSU. There were 18,245 eligible women (ever-married, aged 15–49 years) within the selected households. Of these, 17,863 women were interviewed, yielding a final response rate of 98%. Interviewers recruited in BHDS were well trained; the training includes lectures on how to complete questionnaire, mock interview between participants, and field practice. Data quality standards were maintained through several activities. Four quality control teams, each comprising one male and one female member, from Mitra and Associates were visiting the interviewing teams throughout the data collection period. Two quality control teams from National Institute for Population Research and Training monitor the use of household listing and maps, observed interviewer’s data collection, and spot checked the completed questionnaire. They also ensured whether the selected households were visited and eligible respondents were properly identified and interviewed. Field work was also monitored through visits by the representatives from U.S. Agency for International Development, Monitoring and Evaluation to Access and Use Results Demographic and Health Surveys (MEASURE DHS), and other Technical Review Committee members, which ensured the quality of data collection. Additional details on the data collection procedure were described elsewhere.³³ The analysis discussed here was restricted to women who lived with at least one child younger than 5 years. Participants with missing responses to the variables of interest were dropped from the sample; however, as this number was low (< 1% for most variables), it is unlikely to affect study estimates. The final sample consisted of 7,707 mothers.

Outcomes.

We chose outcome variables that allowed us to evaluate whether children born in the last 5 years had experienced a health concern in a specified timeframe. All outcome-related data were based on maternal report. Infectious disease outcomes included ARI (cough, rapid breathing, blocked, or running nose), diarrhea, or fever in the past 2 weeks. Data were also collected on LBW. However, mothers did not always know the exact birth weights of their children, particularly for those children born at home. Research suggests that subjective assessments of child size are a useful proxy for LBW.³³ For this reason, mothers were asked to estimate child size at birth. Response option included “very small” or “smaller than average.”

Exposures.

Maternal high-risk fertility behaviors were the independent variables of interest. Fertility behaviors were considered “high risk” if identified as such by the 2014 BDHS.³³ The high-risk fertility behaviors examined in this study related to maternal age at the time of delivery, birth order, and birth interval. More specifically, all of the following circumstances were defined as high-risk: mother < 18 years at the time of delivery, mother > 34 years at the time of delivery, most recent child born < 24 months after previous birth, and most recent child of birth order > 3. For analysis, these behaviors were further operationalized as follows: 1) any one high-risk behavior versus none, 2) single high-risk behavior versus none and multiple high-risk behaviors versus none, and 3) each specific behavior versus none.

Covariates.

This study controlled for a variety of socioeconomic characteristics. Children were categorized into five age groups (0–11, 12–23, 24–35, 36–47, or 48–59 months). Owing to the preferential treatment of male children in South Asia (e.g., better nutrition and care), gender of the child was also included in the models.³⁴ Mothers were classified into three age ranges (15–24, 25–34, or 35–49 years) because research suggests that infant and child health varies according to these age groups.³⁵ Maternal educational attainment was categorized as none, primary (1–5 years), secondary, or higher (≥ 6 years). Household socioeconomic status (SES) was derived from the household wealth index reported in the BDHS, which was calculated based on ownership of particular consumer goods and dwelling characteristics.³⁶ Households were then ranked on the basis of their wealth scores and divided into quintiles. Participant religion was categorized as either Muslim or non-Muslim. Given that research suggests maternal autonomy improves child health,³⁷ each mother was assigned an autonomy score on the basis of four questions. The questions asked who in the household makes decisions regarding: maternal health care, large household purchases, visiting family and relatives, and child health care. For each of the four questions, a respondent received 1 point if she was involved in the decision, and 0 point if she was not. These points were summed to yield total scores from 0 to 4 (Cronbach’s alpha = 0.82). Maternal body mass index (BMI) was classified as underweight (< 18.5 kg/m²), normal (18.5–24.99 kg/m²), or overweight/obese (≥ 25 kg/m²). The number of antenatal checkups was dichotomized: fewer than four visits and four or more visits. According to the WHO, four or more visits constitutes sufficient antenatal care.³⁸ A household’s drinking water source was categorized as either protected (e.g., piped water, a protected well) or unprotected. Place of residence was categorized as rural or urban. Previous research has documented regional differences in childhood morbidity in Bangladesh.^7,8 For this reason, we included region (administrative unit) in our analysis.

Statistical analysis.

The sample sociodemographic characteristics were described using weighted percentages. We used modified Poisson regression with generalized estimating equations and robust error variance to determine whether maternal high-risk behaviors predict outcome variables related to childhood morbidity. All models accounted for clustering within PSUs. Unadjusted and adjusted models were fitted for each binary outcome variable (ARI, diarrhea, fever, and LBW) in pairwise combination with each predictor variable (any high-risk behavior versus no risk; separate effects of single high-risk and multiple high-risk versus no risk; specific types of high-risk versus no risk). For both crude and adjusted models, the risk ratios (RRs) were used to assess the strength of the associations, and 95% CIs were computed for significance testing. The cutoff for significance was set at P < 0.05. The final models did not include drinking water source because inclusion of this variable would have resulted in multicollinearity with the SES (source of drinking water was used to construct the SES variable).³⁹ The multicollinearity of the rest of the variables was checked by examining the tolerance values and variance inflation factors (VIFs). In all cases, both tolerance and VIFs were close to unity, indicating no additional issues with multicollinearity.⁴⁰ All statistical analyses were performed in Stata version 13.1/MP (StataCorp, LP, College Station, TX).

RESULTS

Characteristics of the sample.

A total of 7,707 women were eligible for this study. Of these, 3,743 (48.5%) were young, aged 15–24 years. Only 16.4% of women had no education and 74.6% were from rural areas. Nearly a fifth of women (18.5%) did not participate in any household decision-making. From the total sample population, 22.6% of women belonged to the poorest SES and 19.1% belonged to the richest SES. With respect to nutritional status, 59.3% of women were normal weight, 22.2% were undernourished (BMI < 18.5 kg/m²), and 18.5% were overweight or obese (≥ 25 kg/m²). Only 14.3% of women reported four or more antenatal visits during their last pregnancy. Finally, 89.0% of women lived in households with a protected source of drinking water (Table 1).

Table 1.

Sociodemographic characteristics and high-risk fertility behaviors of ever-married women aged 15–49 years who gave birth within the past 5 years, Bangladesh Demographic and Health Survey 2014 (n = 7,707)

Characteristics	n	%*	95% CI
Current age of child (months)
0–11	1,503	19.5	18.5–20.5
12–23	1,624	21.0	19.9–22.3
24–35	1,549	20.1	19.0–21.3
36–47	1,522	19.8	18.6–21.0
48–59	1,509	19.6	18.5–20.7
Gender of child
Male	4,018	52.1	50.5–53.8
Female	3,689	47.9	46.2–49.6
Maternal age (years)
15–24	3,743	48.5	46.7–50.4
25–34	3,379	43.9	42.0–45.7
35–49	585	7.6	6.8–8.5
Maternal education
No education	1,263	16.4	14.4–18.7
Primary	2,156	28.0	26.2–29.8
Secondary+	4,288	55.6	52.7–58.5
Socioeconomic status
Poorest	1,796	22.6	20.0–25.6
Poorer	1,456	18.9	17.4–20.5
Middle	1,496	19.4	17.4–21.6
Richer	1,538	20.0	18.0–22.1
Richest	1,471	19.1	16.7–21.5
Religion
Non-Muslim	654	8.5	6.2–11.5
Muslim	7,053	91.5	88.5–93.8
Decision-making power in household index (Cronbach’s alpha = 0.82)
0 of four items	1,427	18.5	17.0–20.1
1 of four items	940	12.2	11.2–13.3
2 of four items	949	12.3	11.1–13.7
3 of four items	1,006	13.1	11.8–14.4
All four items	3,385	43.9	41.3–46.6
Maternal body mass index†
Underweight	1,713	22.2	20.7–23.9
Normal	4,574	59.3	57.6–61.1
Overweight	1,420	18.5	16.9–20.0
Number of antenatal checkups
< 4	6,603	85.7	84.2–87.0
≥ 4	1,104	14.3	13.0–15.8
Source of drinking water
Protected	6,856	89.0	87.2–90.5
Unprotected	851	11.0	9.5–12.8
Place of residence
Urban	1,957	25.4	23.1–27.9
Rural	5,750	74.6	72.2–77.0
Region
Barisal	441	5.7	4.9–6.7
Chattogram	1,655	21.5	19.2–23.9
Dhaka	2,712	35.2	31.3–39.3
Khulna	580	7.5	6.7–8.4
Rajshahi	791	10.3	9.1–11.5
Rangpur	763	9.9	8.2–11.9
Sylhet	765	9.9	7.7–12.7
Any high-risk fertility behavior
No	5,083	66.0	63.9–68.0
Yes	2,624	34.0	32.0–36.1
Types of high-risk fertility behaviors
No risk	5,083	66.0	63.9–68.0
Single high-risk fertility behaviors	2,210	28.7	27.1–30.1
Mother’s age at birth < 18 years	970	12.6	11.6–13.7
Mother’s age at birth > 34 years	94	1.2	1.0–1.6
Birth interval < 24 months	336	4.4	3.7–5.1
Birth order > 3	810	10.5	9.2–12.0
Multiple high-risk fertility behaviors	414	5.4	4.5–6.3
Age at birth < 18 years and birth interval < 24 months	53	0.7	0.5–1.0
Age at birth < 18 years and birth order > 3	2	0.03	0.0001–0.001
Age < 18 years, birth interval < 24 months, and birth order > 3	–	–	–
Age at birth > 34 years and birth order> 3	231	3.0	0.026–0.034
Age at birth > 34 and birth interval < 24 months	–	–	–
Age at birth > 34 years, birth interval < 24 months, and birth order > 3	27	0.4	0.002–0.005
Birth interval < 24 months and birth order> 3	101	1.3	0.014–0.024

* In estimating percentages, the complex survey design and sampling weights were taken into account.

† Body mass index categories were underweight (< 18.5), normal (18.5–24.9), or overweight/obese (≥ 25).

A substantial proportion of women exhibited at least one high-risk fertility behavior (34.0%); 28.7% engaged in a single high-risk fertility behavior and 5.4% engaged in multiple high-risk fertility behaviors. The most common high-risk fertility behavior was young maternal age at delivery (< 18 years) (12.6%) followed by high birth order (> 3) (10.3%). The most common combination of high-risk fertility behaviors (3.0%) was old maternal age at delivery (> 34 years) and high birth order (> 3) (Table 1).

Prevalence of morbidity and LBW among children.

Table 2 shows the prevalence of morbidity and LBW among children born within the past 5 years. One in 20 (5.4%) children younger than 5 years had experienced ARI in the past 2 weeks; a similar proportion (5.7%) had experienced diarrhea in the same time frame. Approximately one in three (36.9%) children younger than 5 years had experienced fever in the past 2 weeks. Results also indicated that 20.0% of children younger than 5 years were born with an LBW (described by mothers as small or very small at birth).

Table 2.

Prevalence of low birth weight (LBW) and morbidity among children born within the past 5 years to women aged 15–49 years, Bangladesh Demographic and Health Survey 2014 (n = 7,707)

Health indicators	n	%*	95% CI
Acute respiratory infection, past 2 weeks
Yes	416	5.4	4.7–6.1
No	7,291	94.6	93.9–95.4
Diarrhea, past 2 weeks
Yes	439	5.7	5.2–6.2
No	7,268	94.3	93.7–94.8
Fever, past 2 weeks
Yes	2,844	36.9	35.8–38
No	4,863	63.1	62.0–64.2
LBW†
Yes	970	20.0	18.4–21.6
No	3,892	80.0	78.4–81.6

* In estimating percentages, the complex survey design and sampling weights were taken into account.

† Subsample of last born children used for this analysis (n = 4,862).

Bivariate analyses.

Table 3 displays results from the bivariate analyses relating childhood morbidity and LBW to maternal high-risk fertility behaviors and other variables of interest. Engaging in any high-risk fertility behavior was significantly associated with ARI (RR: 1.61, 95% CI: 1.27–2.01), fever (RR: 1.31, 95% CI: 1.06–1.62), and LBW (RR: 1.42, 95% CI: 1.12–1.87). Single high-risk behavior and multiple high-risk behaviors were also associated with ARI, fever, and LBW. Several high-risk behaviors examined separately were significantly associated with childhood morbid conditions. For example, young maternal age at delivery (< 18 years) (RR: 1.48, 95% CI: 1.15–1.91) and old maternal age at delivery (> 34 years) (RR: 1.87, 95% CI: 1.32–2.04) were significantly associated with ARI (Table 3). However, when high-risk fertility behaviors were examined jointly, only the combination of young maternal age at delivery (< 18 years) and short birth interval (< 24 months) was associated with ARI (RR: 1.61, 95% CI: 1.30–2.45). In contrast to high-risk fertility behaviors, maternal education and SES appeared to be protective against ARI, diarrhea, fever, and LBW (Table 3).

Table 3.

Bivariate associations between demographic indicators, high-risk fertility behaviors in women aged 15–49 years, and health outcomes of children younger than 5 years, Bangladesh Demographic Health Survey 2014

Characteristics	Acute respiratory infection	Diarrhea	Fever	Low birth weight*
	Unadjusted risk ratio (95% CI)	Unadjusted risk ratio (95% CI)	Unadjusted risk ratio (95% CI)	Unadjusted risk ratio (95% CI)
Current age of child (months)
0–11 (ref.)	1.00	1.00	1.00	–
12–23	0.80 (0.62–1.04)	1.26 (0.97–1.65)	1.01 (0.92–1.11)	–
24–35	0.80 (0.61–1.05)	0.69 (0.50–0.96)	0.89 (0.82–0.98)	–
36–47	0.55 (0.40–0.74)	0.68 (0.49–0.96)	0.88 (0.82–0.97)	–
48–59	0.46 (0.33–0.65)	0.54 (0.38–0.77))	0.76 (0.69–0.84)	–
Gender of child
Male (ref.)	1.00	1.00	1.00	1.00
Female	0.79 (0.65–0.98)	0.86 (0.71–1.06)	0.97 (0.91–1.03)	1.14 (1.02–1.28)
Maternal age (years)
15–24 (ref)	1.00	1.00	1.00	1.00
25–34	0.71 (0.57–0.87)	0.89 (0.71–1.11)	0.92 (0.86–0.98)	0.95 (0.84–1.08)
35–49	1.07 (0.77–1.47)	0.73 (0.48–1.09)	1.02 (0.92–1.14)	1.14 (0.88–1.47)
Maternal education
No education (ref.)	1.00	1.00	1.00	1.00
Primary	1.32 (0.98–1.77)	1.20 (0.87–1.65)	1.08 (0.98–1.19)	0.79 (0.66–0.94)
Secondary+	0.78 (0.63–0.94)	0.90 (0.66–0.98)	0.98 (0.89–1.08)	0.62 (0.53–0.73)
Socioeconomic status
Poorest (ref.)	1.00	1.00	1.00	1.00
Poorer	0.95 (0.71–1.26)	1.06 (0.78–1.45)	0.93 (0.84–1.04)	0.81 (0.68–0.98)
Middle	0.90 (0.67–1.21)	0.83 (0.59–1.17)	1.03 (0.93–1.14)	0.82 (0.68–0.98)
Richer	0.74 (0.55–0.98)	0.80 (0.56–1.13)	0.94 (0.85–1.01)	0.70 (0.58–0.84)
Richest	0.54 (0.38–0.76)	0.86 (0.61–1.21)	0.82 (0.73–0.93)	0.64 (0.52–0.78)
Religion
Non-Muslim (ref.)	1.00	1.00	1.00	1
Muslim	1.34 (0.89–2.01)	1.04 (0.68–1.59)	1.17 (1.02–1.34)	0.93 (0.76–1.56)
Decision-making power in household index (Cronbach’s alpha = 0.82)
0 of four items	1.00	1.00	1.00	1.00
1 of four items	0.82 (0.59–1.13)	0.91 (0.64–1.31)	1.03 (0.92–1.15)	1.12 (0.91–1.36)
2 of four items	0.89 (0.64–1.25)	0.91 (0.64–1.28)	1.04 (0.93–1.15)	0.88 (0.71–1.09)
3 of four items	0.83 (0.65–0.98)	0.82 (0.57–1.18)	1.01 (0.91–1.12)	0.97 (0.80–1.20)
All four items	0.74 (0.57–0.83)	0.83 (0.62–1.11)	0.89 (0.82–0.98)	0.87 (0.74–1.03)
Maternal body mass index†
Underweight (ref.)	1.00	1.00	1.00	1.00
Normal weight	0.68 (0.55–0.86)	0.84 (0.66–1.07)	0.91 (0.84–0.97)	0.75 (0.66–0.86)
Overweight	0.54 (0.39–0.74)	0.68 (0.48–0.98)	0.80 (0.72–0.88)	0.64 (0.52–0.78)
Number of antenatal checkups
< 4 (ref.)	1.00	1.00	1.00	1.00
≥ 4	0.95 (0.72–1.26)	0.95 (0.72–1.25)	1.10 (1.01–1.19)	0.88 (0.77–1.01)
Source of drinking water
Protected (ref.)	1.00	1.00	1.00	1.00
Unprotected	1.34 (1.10–1.58)	1.22 (1.01–1.66)	0.95 (0.85–1.06)	1.17 (0.99–1.30)
Place of residence
Urban (ref.)	1.00	1.00	1.00	1.00
Rural	1.29 (1.01–1.67)	0.98 (0.77–1.24)	1.05 (0.98–1.14)	1.17 (1.02–1.35)
Region
Barisal (ref.)	1.00	1.00	1.00	1.00
Chattogram	1.32 (0.88–1.97)	1.13 (0.75–1.72)	0.98 (0.87–1.11)	1.30 (1.01–1.69)
Dhaka	1.23 (0.80–1.88)	0.90 (0.58–1.39)	0.88 (0.78–0.99)	1.36 (1.05–1.76)
Khulna	1.55 (1.10–2.01)	0.63 (0.38–0.95)	0.84 (0.73–0.97)	1.11 (0.83–1.48)
Rajshahi	1.50 (0.98–2.29)	0.72 (0.42–1.24)	0.97 (0.84–1.12)	0.93 (0.70–1.24)
Rangpur	1.34 (0.88–2.04)	0.46 (0.28–0.78)	0.99 (0.87–1.15)	0.95 (0.71–1.27)
Sylhet	1.48 (1.02–2.25)	1.02 (0.64–1.61)	1.03 (0.91–1.18)	1.68 (1.32–2.15)
Any high-risk category
No (ref.)	1.00	1.00	1.00	1.00
Yes	1.61 (1.27–2.01)	1.09 (0.89–1.34)	1.31 (1.06–1.62)	1.42 (1.12–1.87)
Types of high-risk category
No risk (ref.)	1.00	1.00	1.00	1.00
Single high-risk category	1.44 (1.08–1.65)	1.31 (0.91–1.40)	1.32 (1.09–1.62)	1.21 (1.07–1.37)
Multiple high-risk category	1.32 (1.04–1.56)	0.93 (0.58–1.50)	1.26 (1.03–1.41)	1.31 (1.04–1.66)
Specific high-risk category
No risk	1.00	1.00	1.00	1.00
Mother’s age at delivery < 18 years	1.48 (1.15–1.91)	1.27 (1.01–1.67)	1.09 (0.99–1.19)	1.24 (1.05–1.47)
Mother’s age at delivery > 34 years	1.87 (1.32–2.04)	1.27 (0.58–2.76)	1.15 (0.90–1.47)	1.45 (1.11–1.85)
Birth interval < 24 months	0.98 (0.57–1.66)	1.01 (0.59–1.69)	1.14 (0.98–1.31)	1.06 (0.78–1.43)
Birth order > 3	1.23 (0.91–1.68)	1.12 (0.71–1.36)	1.16 (1.06–1.27)	1.25 (1.04–1.50)
Age at delivery < 18 years and birth interval < 24 months‡	1.61 (1.30–2.45)	0.82 (0.45–3.09)	1.12 (0.78–1.60)	0.44 (0.12–1.68)
Age at delivery > 34 years and birth interval < 24 months and birth order > 3§	1.07 (0.61–1.88)	0.92 (0.46–1.79)	1.08 (0.92–1.27)	1.50 (1.15–1.96)
Birth interval < 24 months and birth order > 3	1.24 (0.66–2.33)	1.02 (0.78–2.01)	1.31 (1.10–1.56)	1.18 (0.72–1.90)

* Subsample of last born children used for this analysis (n = 4,862).

† Body mass index categories were underweight (< 18.5), normal (18.5–24.9), or overweight/obese (≥ 25.0).

‡ Includes the categories: age at delivery < 18 years; birth interval < 24 months, age < 18, and birth order > 3.

§ Includes the categories: age at delivery > 34 years and birth order > 3; age at delivery > 34 years and birth interval < 24 months; and age at delivery > 34 years, birth interval < 24 months, and birth order > 3.

Multivariate analyses.

After adjusting for relevant covariates, high-risk fertility behaviors increased the risk of ARI, diarrhea, fever, and LBW by 22%, 18%, 29%, and 27%, respectively. Compared with children of women who were not engaged in any high-risk behaviors, children of women who engaged in a single high-risk fertility behavior were 1.24 times more likely to experience an ARI in the past 2 weeks (ARR: 1.24, 95% CI: 1.09–1.56), 1.21 times more likely to experience diarrhea in the past 2 weeks (ARR: 1.21, 95% CI: 1.07–1.39), and 1.15 times more likely to have been of LBW (ARR: 1.15, 95% CI: 1.01–1.33) (Table 4). A significant association was observed between engaging in multiple high-risk fertility behaviors and having a child who experienced ARI (ARR: 1.91, 95% CI: 1.40–2.72), diarrhea (ARR: 1.93, 95% CI: 1.42–2.72), or fever (ARR: 1.09, 95% CI: 1.03–1.25) within the past 2 weeks as well as having had an LBW infant (ARR: 1.27, 95% CI: 1.07–1.45). In terms of specific high-risk fertility behaviors, younger maternal age at delivery (< 18 years) was significantly associated with childhood ARI, diarrhea, and having had an LBW infant. Among the combinations of high-risk fertility behaviors, shorter birth interval (< 24 months) and higher birth order (> 3) were significantly associated with having a child experiencing ARI (ARR: 1.44, 95% CI: 1.24–2.11), diarrhea (ARR: 1.91, 95% CI: 1.51–2.85), or fever (ARR: 1.33, 95% CI: 1.11–1.58) as well as having been born with an LBW (ARR: 2.29, 95% CI: 1.65–3.45) (Table 4).

Table 4.

Adjusted risk ratios (ARR) for associations between high-risk fertility behaviors in women aged 15–49 years and health outcomes of children younger than 5 years, Bangladesh Demographic and Health Survey 2014

Measure	Acute respiratory infection	Diarrhea	Fever	Low birth weight
	ARR* (95% CI)	ARR* (95% CI)	ARR* (95% CI)	ARR† (95% CI)
Any high-risk category
No	1.00	1.00	1.00	1.00
Yes	1.22 (1.05–1.50)	1.18 (1.03–1.35)	1.29 (1.11–1.58)	1.27 (1.10–1.52)
Types of high-risk category
No risk	1.00	1.00	1.00	1.00
Single high-risk category	1.24 (1.09–1.56)	1.21 (1.07–1.39)	1.25 (0.96–1.20)	1.15 (1.01–1.33)
Multiple high-risk category	1.91 (1.40–2.72)	1.93 (1.42–2.72)	1.09 (1.03–1.25)	1.27 (1.07–1.45)
Specific high-risk category
No risk	1.00	1.00	1.00	1.00
Mother’s age at delivery < 18 years	1.24 (1.07–1.71 )	1.26 (1.05–1.74)	1.01 (0.95–1.15)	1.18 (1.01–1.41)
Mother’s age at delivery > 34 years	1.30 (0.57–3.01)	1.98 (0.72–5.70)	1.11 (0.82–1.45)	1.19 (0.89–1.61)
Birth interval < 24 months	1.14 (0.59–1.58)	0.95 (0.52–1.51)	1.47 (0.90–2.41)	0.97 (0.71–1.31)
Birth order > 3	1.16 (0.80–1.61)	0.86 (0.61–1.20)	1.18 (1.10–1.31)	1.01 (0.83–1.00)
Age at delivery < 18 years and birth interval < 24 months‡	1.32 (0.58–3.29)	0.95 (0.34–3.53)	2.27 (1.24–4.03)	0.35 (0.09–1.34)
Age at delivery > 34 years and birth interval < 24 months and birth order > 3§	0.65 (0.31–1.47)	1.19 (0.49–3.18)	1.01 (0.81–1.32)	1.26 (0.85–1.87)
Birth interval < 24 months and birth order > 3	1.44 (1.24–2.11)	1.91 (1.51–2.85)	1.33 (1.11–1.58)	2.29 (1.65–3.45)

ARR = adjusted relative risk.

* Adjusted for age of child, gender of child, maternal age, education, socioeconomic status (SES), religion, decision-making power, maternal BMI, number of antenatal care (ANC) visit, place of residence, and region.

† Subsample of last born children used for this analysis (n = 4,862) and adjusted for age and gender of child, maternal age, education, SES, religion, decision-making power, maternal BMI, number of ANC visit, place of residence, and region.

‡ Includes the categories: age at delivery < 18 years; birth interval < 24 months, age < 18, and birth order > 3.

§ Includes the categories: age at delivery > 34 years and birth order> 3; age at delivery > 34 years and birth interval < 24 months; and age at delivery > 34 years, birth interval < 24 months, and birth order > 3.

DISCUSSION

To the best of our knowledge, this study is the first to assess the impact that a combination of maternal high-risk fertility behaviors has on the risk of childhood morbidity (ARI, diarrhea, and fever) and LBW. Our findings show that children born to women who engage in high-risk fertility behaviors are significantly more vulnerable to illness and LBW. Moreover, given that these data come from a nationally representative survey, we were able to determine that these practices are shockingly routine: more than a third of Bangladeshi women (34.0%) reported at least one high-risk behavior. Furthermore, despite improvements in child health over the past decade, our findings reveal that childhood ARI (5.4%), diarrhea (5.7%), and fever (36.9%) as well as LBW (20.0%) are still major causes of concern in Bangladesh.

This study revealed that the chances of ARI, diarrhea, fever, and LBW were higher in children whose mothers engaged in at least one high-risk fertility behavior. These associations suggest that high-risk fertility behaviors threaten the health and well-being of children. Moreover, these effects persist when adjusting for relevant demographic, socioeconomic, and maternal and child factors known to predict child health and LBW. Overall, our findings regarding the association between high-risk fertility behavior, childhood morbidity, and LBW are in line with previous research demonstrating that a range of negative health consequences stem from maternal high-risk fertility behaviors.^{9,15,21,41–46}

Given our use of control variables, it seems clear that these associations are not simply a consequence of maternal socioeconomic vulnerability or maternal malnutrition (as indicated by low BMI). Thus, additional research is needed to investigate why childhood morbidity and LBW are more common in children born to mothers who engage in high-risk fertility behaviors. Previous studies suggest that women who get married before 18 years are more controlled by their husbands and in-laws,^47–49 and this might limit their access to sufficient food for themselves and their children. Insufficient access to food, combined with the limited nutritional reserves stored within the bodies of adolescent mothers, is liable to place offspring at substantial risk for LBW and insufficient access to breast milk.^50,51 The association between young maternal age at delivery (< 18 years) and LBW suggests that the effects of inadequate fetal nutrition and reduced breastfeeding among neonates born to adolescent mothers extend to infancy and early childhood, heightening the risk for later illness.^52,53 Early childbearing may also affect child health indirectly. For example, childbearing adversely affects educational attainment; girls in developing countries often drop out of school or college following marriage.^54,55 Furthermore, low levels of education generally lead to lower income and low autonomy, which together limit nutritional purchasing power.⁵⁶ Thus, improving child health will require interventions to delay marriage until after a woman is 18 years old.

Our findings also suggest that older maternal age at delivery (> 34 years) increases the risk of LBW. Earlier research has found that advanced maternal age predicts a gradual deterioration of the intrauterine environment as well as decreased viability of embryos due to the age-dependent decreases in oocyte quality.⁵⁷ These changes mean that older mothers are at higher risk of pregnancy complications. In the same vein, the risk of miscarriage, preterm birth, LBW, still birth, and Down syndrome increase exponentially with age.^58,59 In addition, older mothers are known to suffer more from hypertension, preeclampsia, and gestational diabetes, which are all diseases associated with one’s risk of having an LBW infant.^60–62

Furthermore, risk factors such as a failure to use health-care services and socioeconomic disadvantages are more commonly observed among women who exhibit short interpregnancy intervals, a higher number of total live births, and an early or late childbearing pattern.⁶³ These risk factors might explain the link between high-risk fertility behaviors, child morbidity, and LBW. Another important finding of our study was that engaging in multiple high-risk fertility behaviors had a stronger impact on child health than engaging in only one high-risk fertility behavior. More specifically, we observed a dose–response relationship between the types of high-risk fertility behaviors and childhood morbidity and LBW. Previous research suggests that engaging in multiple high-risk fertility behaviors predicts higher levels of depressive symptoms, a higher chance of receiving inappropriate preventive and curative services, and an increased likelihood of inadequate breast-feeding practices.^64,65 Each of these factors may contribute to poor child health outcomes.

The main strength of this study was that the data came from a large, nationally representative survey. Moreover, our findings offer important insights into maternal high-risk fertility behaviors and their effects on child morbidity in Bangladesh. Nonetheless, these results must be interpreted in the light of several limitations. The outcomes used in this study were based on self-report, which leave responses vulnerable to social desirability and recall biases. Indeed, in their review of 110 Demographic and Health Surveys, Manesh et al.⁶⁶ reported that recall and reporting biases related to under-5 morbidity were likely significant; therefore, caution should be exercised when using and interpreting child health data from a DHS. In addition, because this study was cross-sectional, causality cannot be assumed. For example, it is difficult to account for seasonal differences in the occurrences of illnesses such as ARI, diarrhea, and fever. Notably, however, BDHS data were collected over a 6-month period covering both wet and dry seasons, and this may ultimately have minimized any seasonal effects. Still, prospective studies will improve our understanding of the effects that high-risk fertility behaviors have on child morbidity. This study’s another limitation was its use of maternal estimates of infant size instead of actual birth weight data; this measure has questionable reliability and validity.

CONCLUSION

In spite of these limitations, this study offers important information that could help reduce child morbidity and mortality in Bangladesh. Maternal high-risk behaviors are shockingly frequent among Bangladeshi women, and these behaviors were important predictors of ARI, diarrhea, fever, and LBW. Thus, it appears that efforts to reduce high-risk fertility behaviors will help also combat childhood morbidity in Bangladesh. Moreover, although not explicitly tested, these study findings may be relevant in other resource-limited settings where ARI, diarrhea, fever, and LBW are common. These findings may help policy makers, development partners, and other relevant institutions develop appropriate intervention programs. Future studies capable of determining whether there is a causal link between high-risk fertility behaviors and child morbidity should be a research priority.