BDDHS preliminary data analysis

Getting started

Here we show the pre-requisite code sections. Run these at the outset to avoid errors. First we load the required packages.

easypackages::libraries(
    
    # Data i/o
    "here",                 # relative file path
    "rio",                  # file import-export
    
    # Data manipulation
    "janitor",              # data cleaning fns
    "haven",                # stata, sas, spss data io
    "labelled",             # var labelling
    "readxl",               # excel sheets
    # "scales",               # to change formats and units
    "skimr",                # quick data summary
    "broom",                # view model results
    
    # Data analysis
    "DHS.rates",            # demographic rates for dhs-like surveys
    "GeneralOaxaca",        # BO decomposition for non-linear
    "survey",               # apply survey weights
    "survival",             # base pkg for survival analysis
    
    # Analysis output
    "gt",
    # "modelsummary",          # output summary tables
    "gtsummary",            # output summary tables
    "flextable",            # creating tables from objects
    "officer",              # editing in office docs
    
    # R graph related packages
    "ggstats",              # graph of regression estimates
    "RColorBrewer",         # graph themes
    # "scales",
    "patchwork",            # combining graphs
    
    # Misc packages
    "tidyverse",            # Data manipulation iron man
    "tictoc"                # Code timing
)

Next we turn off scientific notations.

options(scipen = 999)

Next we set the default gtsummary print engine for tables.

theme_gtsummary_printer(print_engine = "flextable")

Now we set the flextable output defaults.

set_flextable_defaults(
  font.size = 11,
  text.align = "left",
  big.mark = "",
  background.color = "white",
  table.layout = "autofit",
  theme_fun = theme_vanilla
)

Document introduction

In this document we do preliminary analysis of examining infant mortality clustering under mothers in the pooled Bangladesh Demographic and Health Survey (BDDHS) datasets. This document has selected code for analysis and the full spectrum of preliminary analysis are available in the “bddhs_datanal1.R” script file.

Data source

Here we use the pooled dataset of 9 existing rounds of Bangladesh Demographic and Health Survey (BDDHS) conducted during 1993, 1996, 1999, 2004, 2007, 2011, 2014, 2017 and 2022. The BDDHS implemented a multi-stage cluster sampling design. Details regarding the sample design, survey instruments, informed consent, training of interviewers and other fieldwork officers, data collection and processing, and response rates are available in the final report of the respective surveys.

We pooled the raw birth histories of ever-married women in the reproductive age group (15-49 years) across all the pre-specified rounds. We manually harmonized all required datasets so that our study variables have consistent categorisation and label codes across the surveys. The pooled dataset harmonization and pooling codes are in “bddhsv2_daprep.R”. Also, we have a variable harmonization documentation called “bddhs_datapl1_pre.html” which shows how the variable labels and label codes vary across the raw BDDHS datasets.

Now, please note that the following Bangladesh DHS surveys were pooled:

# Creating the table of surveys to be used for pooling
bdbr1_tmp_intro |> 
  mutate(n_births = prettyNum(n_births, big.mark = ",")) |> 
  select(c(ctr_name, svy_year, n_births)) |> 
    # Join vars from bdhr_tmp_intro
    left_join(
        bdhr1_tmp_intro |> 
          mutate(n_households = prettyNum(n_households, big.mark = ",")) |> 
          select(svy_year, n_households),
        by = join_by(svy_year)
    ) |> 
    # Join vars from bdpr_tmp_intro
    left_join(
        bdpr1_tmp_intro |> 
          mutate(n_persons = prettyNum(n_persons, big.mark = ",")) |> 
          select(svy_year, n_persons),
        by = join_by(svy_year)
    ) |> 
    # convert nested tibble to simple tibble
    unnest(cols = c()) |> 
    mutate(
        ccode = row_number(), 
        .before = ctr_name
    ) |> 
    # convert to flextable object
    qflextable() |> 
    theme_vanilla() |> 
    align(align = "left", part = "all")

Table 1: Bangladesh DHS datasets and their sample size to be used for pooling

ccode	ctr_name	svy_year	n_births	n_households	n_persons
1	Bangladesh	1993	32,581	9,174	51,631
2	Bangladesh	1996	29,344	8,682	47,432
3	Bangladesh	1999	31,906	9,854	54,627
4	Bangladesh	2004	33,597	10,500	55,883
5	Bangladesh	2007	30,527	10,400	53,413
6	Bangladesh	2011	45,833	17,141	83,731
7	Bangladesh	2014	43,772	17,300	81,624
8	Bangladesh	2017	47,828	19,457	89,819
9	Bangladesh	2022	64,722	30,018	132,463

Note that the information on birth histories was available for 164,761 children. The current study utilised pooled retrospective birth histories of children born between 1970 to 2022 (till the interview date). Therefore, the analytical sample size of this study is 163,544 children.

Study variables

We use the following variables for the pooled data analysis:

• Dependent variable
  • infantd = Index child died during infancy period (0-11 months)
• Main Independent variable
  • sibsurv_nmv = Survival status of preceding child (Death scarring)
  • binterval_3c_nmv_opp = Birth interval preceding to index child
• Independent variables [CHILD LEVEL]
  • cyob10y_opp = Birth cohort of index child
  • bord_c = Birth order of index child
  • sex_fm = Gender of index child
  • season = Season during birth
• Independent variables [MOTHER/PARENT LEVEL]
  • myob_opp = Birth cohort of mother
  • macb_c_opp = Mother’s age during birth of index child
  • medu_opp = Mother’s Level of education
  • fedu_opp = Father’s level of education
• Independent variables [HOUSEHOLD LEVEL]
  • religion = Religion
  • nat_lang = Native language of respondent
  • wi_qt_opp = Household wealth quintile
  • hhgen_2c_opp = Generations in household
  • hhstruc_opp = Household structure
  • head_sex_fm = Sex of HH head
• Independent variables [COMMUNITY LEVEL]
  • por = Place of residence of the household
  • ecoreg = Ecological region

Note: (a) Crossed names indicates variable not included.

Statistical methods

We first conducted descriptive analysis to examine the absolute and percentage distribution of births by background characteristics in the pooled dataset. Next we explored the distribution of infant mortality by birth cohorts and survey periods. Also, we examined the trend in childhood mortality (neonatal, infant and under-five mortality) rates across the Bangladesh DHS survey rounds. Next, we performed bivariate analysis to assess the unadjusted relationship between infant mortality and the main independent variables using the chi-square test for association. To estimate the adjusted associations, we applied multivariable logistic regression models on the pooled dataset covering 1970-2022.

Data import

We will directly import the prepared dataset for analysis here. The code for dataset preparation is in the “bddhsv2_daprep.R” script file.

# Importing the pooled bdbr tibble
bdbr1_pl1 <- read_rds(
    file = here("website_data", "bdbr1_pl1.rds")
)

# Prepare the survey weighted dataset
options(survey.lonely.psu = "adjust")
options(survey.adjust.domain.lonely = T)
svy_bdbr1_pl1 <- svydesign(
    data = bdbr1_pl1,
    ids = ~ as.numeric(svy_psu),
    strata = ~ as.numeric(svy_strata),
    weight = ~ svy_wwgt6,
    probs = NULL,
    nest = T
)

Descriptive analysis

Sample distribution of births

We will check the distribution of births by background characteristics in Bangladesh 1970-2022.

# Now, we merge the unweighted and weighted tables and view it
t1 <- tbl_merge(
    tbls = list(t1_wt, t1_ut),
    tab_spanner = c("**Weighted**", "**Unweighted**")
) |> 
    modify_footnote(
        stat_0_1 ~ "n (%) = Distribution and column percentage (weighted);",
        stat_0_2 ~ "n (%) = Distribution and column percentage (unweighted)"
    ) |> 
  as_flex_table() |> 
  theme_vanilla()
t1

Table 2: Distribution of births by background characteristics in Bangladesh 1970-2022

	Weighted	Unweighted
Characteristic	N = 913,346,5561	N = 349,9622
Death scarring
Alive	514,904,887 (56.4%)	195,665 (55.9%)
Dead	97,293,820 (10.7%)	34,379 (9.8%)
First order births	301,147,850 (33.0%)	119,918 (34.3%)
Death scarring
Alive	514,904,887 (56.4%)	195,665 (55.9%)
Dead after conception	23,980,882 (2.6%)	8,239 (2.4%)
Dead before conception	73,312,937 (8.0%)	26,140 (7.5%)
First order births	301,147,850 (33.0%)	119,918 (34.3%)
Preceding birth interval (in months)
24 and more months	472,154,835 (51.7%)	177,834 (50.8%)
18-23 months	79,264,465 (8.7%)	29,220 (8.3%)
Less than 18 months	60,779,406 (6.7%)	22,990 (6.6%)
First order births	301,147,850 (33.0%)	119,918 (34.3%)
Birth cohort of index child
2010 and above	107,107,560 (11.7%)	55,742 (15.9%)
2000-2009	214,936,478 (23.5%)	92,065 (26.3%)
1990-1999	287,372,873 (31.5%)	103,624 (29.6%)
1980-1989	220,106,757 (24.1%)	72,189 (20.6%)
1970-1979	83,822,888 (9.2%)	26,342 (7.5%)
Birth order
1-2	536,207,139 (58.7%)	212,503 (60.7%)
3	153,898,955 (16.9%)	58,420 (16.7%)
4	93,841,570 (10.3%)	34,252 (9.8%)
5 and more	129,398,892 (14.2%)	44,787 (12.8%)
Sex of child
Female	445,492,696 (48.8%)	170,525 (48.7%)
Male	467,853,861 (51.2%)	179,437 (51.3%)
Season during birth
Summer/pre-monsoon	272,262,473 (29.8%)	104,399 (29.8%)
Monsoon	208,789,167 (22.9%)	80,232 (22.9%)
Winter/post-monsoon	432,294,917 (47.3%)	165,331 (47.2%)
Mother's age at child birth (in years)
Above 29 years	103,048,119 (11.3%)	39,265 (11.2%)
25-29 years	169,767,336 (18.6%)	66,132 (18.9%)
20-24 years	292,464,253 (32.0%)	113,470 (32.4%)
Below 20 years	348,066,848 (38.1%)	131,095 (37.5%)
Mother's education
Secondary or higher	232,945,635 (25.5%)	102,885 (29.4%)
Upto Primary	278,431,181 (30.5%)	110,140 (31.5%)
No formal schooling	401,969,740 (44.0%)	136,937 (39.1%)
Father's education
Secondary or higher	278,004,208 (30.4%)	110,433 (31.6%)
Upto primary	237,404,080 (26.0%)	90,147 (25.8%)
No formal schooling	397,938,269 (43.6%)	149,382 (42.7%)
Religion of the household
Islam	827,755,878 (90.6%)	315,698 (90.2%)
Others	85,503,455 (9.4%)	34,238 (9.8%)
Unknown	87,223	26
Wealth index quintile
Richest	157,253,196 (17.2%)	68,327 (19.5%)
Richer	177,418,951 (19.4%)	68,309 (19.5%)
Middle	185,811,445 (20.3%)	69,353 (19.8%)
Poorer	195,948,801 (21.5%)	71,723 (20.5%)
Poorest	196,914,163 (21.6%)	72,250 (20.6%)
Household structure
Extended	430,842,462 (47.2%)	167,753 (47.9%)
Nuclear	482,504,094 (52.8%)	182,209 (52.1%)
Sex of household head
Female	96,735,517 (10.6%)	38,193 (10.9%)
Male	816,611,039 (89.4%)	311,769 (89.1%)
Place of residence
Urban	182,944,415 (20.0%)	100,837 (28.8%)
Rural	730,402,142 (80.0%)	249,125 (71.2%)
1n (%) = Distribution and column percentage (weighted);
2n (%) = Distribution and column percentage (unweighted)

Table 2 presents the sample characteristics of all births from the complete birth histories of Bangladesh from 1970 to 2022. The distribution of our two main explanatory variables, shows that there were 11% children whose previous sibling had died before their birth. Most children in Bangladesh had a preceding birth interval of 24 months or more. Among the children considered in the study, the majority were from the 1990-99 birth cohort (35.5%), followed by the 2000-09 cohort (23.1%) and the 1980-89 cohort (22.7%). More than half of the children were male (51.2%) and belonged to the first or second birth order (56.4%).

Almost 38% of children were born when their mothers were aged 20-24 years, followed by 24% whose mothers were less than 20 years old. Nearly 71% of mothers and 36% of fathers had no formal schooling. Most households practised Hinduism (84.8%), and Bangladeshi was the primary language spoken in almost 46% of households. The majority of children lived in rural areas (75.7%) and in the Terai region (52.7%), followed by the Hill region (40%) and the Mountain region (7.3%).

Distribution of infant deaths by birth cohort and survey period

Here we prepare the data of infant mortality by birth cohort and then by survey period. Using the data we prepare the graph of distribution of infant deaths by birth cohort and survey period.

# Let's create the lollipop chart.
fig1_cohort <- f1_cohort_tibble |> 
    ggplot(aes(y = birth_cohort, x = pct_dead)) +
    
    # adding line segment or lollipop stick
    geom_segment(
        aes(x = 0, xend = pct_dead, y = birth_cohort, yend = birth_cohort),
        linewidth = 2,
        colour = "green4"
    ) +
    
    # adding the point or lollipop head
    geom_point(
        shape = 21,
        size = 15,
        fill = "white",
        colour = "green4"
    ) +
    
    # adding the percentage text
    geom_text(
        aes(label = pct_dead),
        fontface = "bold"
    ) +
    
    labs(
        # title = "(A) By Birth cohort",
        x = "Weighted percentage of children",
        y = "Birth cohort",
    ) +
    
    theme(
        plot.title = element_text(size = rel(1.3), face = "bold"),
        # plot.subtitle = element_text(size = rel(1.2)),
        axis.title.x = element_text(size = rel(1.2)),
        axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        axis.title.y = element_text(size = rel(1.2)),
        axis.text.y = element_text(size = rel(1.2))
    )
fig1_cohort

Figure 1: Lollipop chart showing the weighted percentage of infant deaths by birth cohort in Bangladesh DHS

fig1_period <- f1_period_tibble |> 
    ggplot(aes(x = pct_dead, y = survey_year)) +
    
    # adding the line
    geom_segment(
        aes(x = 0, xend = pct_dead, y = survey_year, yend = survey_year),
        linewidth = 3,
        colour = "orange3"
    ) +
    
    # adding the point
    geom_point(
        shape = 21,
        size = 18,
        fill = "white",
        colour = "orange3"
    ) +
    
    # adding the percentage label
    geom_text(
        aes(label = pct_dead),
        fontface = "bold"
    ) +
    
    labs(
        # title = "(B) By Survey period",
        x = "Weighted percentage of children",
        y = "Survey period",
    ) +
    
    theme(
        plot.title = element_text(size = rel(1.3), face = "bold"),
        axis.title.x = element_text(size = rel(1.2)),
        axis.text.x = element_blank(),
        axis.ticks.x = element_blank(),
        axis.title.y = element_text(size = rel(1.2)),
        axis.text.y = element_text(size = rel(1.2))
    )
fig1_period

Figure 2: Lollipop chart showing the weighted percentage of infant deaths by survey period in Bangladesh DHS

Figure 1 and Figure 2 further highlights the temporal shifts in infant mortality across different birth cohorts and survey periods. The results indicate a steady decline in infant mortality over time, with higher mortality rates in earlier birth cohorts (1970–1979), which gradually decreased in the subsequent cohorts. The most recent cohorts (2010 and later) exhibit the lowest mortality rates, reflecting improvements in child survival over the decades.

A similar trend can be observed for infant mortality in the survey rounds. Proportion of children who died during infancy were higher during the 1996 round, which continued to decrease in the recent 2022 round.

Childhood mortality rates by survey rounds

Here we calculate mortality rates across the Bangladesh DHS survey rounds using the DHS.Rates R package. First we prepared the dataset of childhood mortality rates and then we drew the graph.

# Now we prepare the connected scatter plots of nnmr, imr and u5mr
fig2_svyround <- bdbrpl1_mortrates |> 
    # filter the mortality rates
    filter(mort_stat %in% c("nnmr", "imr", "u5mr")) |> 
    # prepare the graph
    ggplot(aes(x = year, y = r, colour = mort_stat)) +
    # add line
    geom_path(linewidth = 1, alpha = 0.9) +
    # add point
    geom_point(size = 16, alpha = 0.9) +
    # add label
    geom_text(
        aes(label = round(r, 0)), 
        colour = "white",
        fontface = "bold"
    ) +
    # give breaks according to survey round
    scale_x_continuous(breaks = c(1996, 2001, 2006, 2011, 2016, 2022)) +
    scale_colour_discrete(
        label = c("Neonatal mortality rate", "Infant mortality rate", 
                  "Under-five mortality rate")
    ) +
    
    # add labels
    labs(
        title = "Trends in Childhood mortality rates in Bangladesh DHS survey rounds",
        x = "Survey round",
        y = "Mortality rate (per 1000 live births)",
        # This removes the legend title 
        colour = NULL
    ) +
    # Modify the axis theme elements
    theme(
        plot.title = element_text(size = rel(1.4), face = "bold"),
        axis.title.x = element_text(size = rel(1.2)),
        axis.text.x = element_text(size = rel(1.2)),
        axis.title.y = element_text(size = rel(1.2)),
        axis.text.y = element_blank(),
        axis.ticks.y = element_blank()
    ) +
    # Modify the legend theme elements
    theme(
        # legend.key.size = unit(1, "points"),
        legend.position = "top",
        legend.text = element_text(size = rel(0.9))
    ) +
    guides(color = guide_legend(override.aes = list(size = 8)))
fig2_svyround    

Figure 3: Connected dot plot showing the childhood mortality rates by Bangladesh DHS survey rounds

Figure 3 also shows the childhood mortality rates (per 1000 live births) have decreased across the survey rounds. This pattern stands true for neonatal, infant and under-five mortalities. Additionally, this trend corroborates with the declining infant mortality trend observed in Figure 1 and Figure 2.

Bivariate analysis

Crosstab of infant mortality by background characteristics

Here we will check the crosstab tables of infant mortality with the independent variables using row percentages and check their association with the chi-square test.

# Creating the var list
vlist_t2 <- c("infantd", "sibsurv_3c", "binterval_3c_opp", 
              "cyob10y_opp", "bord_c", "sex_fm", "season", "macb_c_opp", 
              "medu_opp", "pedu_opp", "religion", "wi_qt_opp", 
              "hhstruc_opp", "head_sex_fm", "por")

# Prepare the weighted gtsummary table
t2_wt <- svy_bdbr1_pl1 |> 
  tbl_svysummary(
    include = all_of(vlist_t2),
    by = infantd,
    percent = "row"
  ) |> 
  add_overall(statistic = ~ "{n}") |> 
  add_p() |> 
  
  # hide the alive col
  modify_column_hide(columns = stat_1) |> 
  # Add extra header
  modify_spanning_header(
    c(stat_2, p.value) ~ "**Infant mortality (0-11 months)**"
  ) |> 
  modify_footnote(
    all_stat_cols() ~ "n = Absolute distribution; % = Row percentage; Adjusted for sampling weights"
  ) |> 
  # styling
  bold_labels() |> 
  bold_p(t = 0.05) |> 
  as_flex_table() |> 
  theme_vanilla()
t2_wt                           # view table

Table 3: Distribution of infant deaths by background characteristics among Bangladeshi children from 1970-2022

		Infant mortality (0-11 months)
Characteristic	Overall N = 913,346,5561	Dead N = 71,865,2861	p-value2
Death scarring			<0.001
Alive	514,904,887	29,689,480 (5.8%)
Dead after conception	23,980,882	2,805,700 (12%)
Dead before conception	73,312,937	10,190,574 (14%)
Unknown	301,147,850	29,179,531
Preceding birth interval (in months)			<0.001
24 and more months	472,154,835	25,846,519 (5.5%)
18-23 months	79,264,465	7,880,063 (9.9%)
Less than 18 months	60,779,406	8,959,173 (15%)
Unknown	301,147,850	29,179,531
Birth cohort of index child			<0.001
2010 and above	107,107,560	3,529,964 (3.3%)
2000-2009	214,936,478	10,378,952 (4.8%)
1990-1999	287,372,873	22,005,146 (7.7%)
1980-1989	220,106,757	24,084,401 (11%)
1970-1979	83,822,888	11,866,824 (14%)
Birth order			<0.001
1-2	536,207,139	44,958,994 (8.4%)
3	153,898,955	9,723,936 (6.3%)
4	93,841,570	6,421,801 (6.8%)
5 and more	129,398,892	10,760,555 (8.3%)
Sex of child			<0.001
Female	445,492,696	31,850,438 (7.1%)
Male	467,853,861	40,014,848 (8.6%)
Season during birth			0.8
Summer/pre-monsoon	272,262,473	21,276,399 (7.8%)
Monsoon	208,789,167	16,434,022 (7.9%)
Winter/post-monsoon	432,294,917	34,154,865 (7.9%)
Mother's age at child birth (in years)			<0.001
Above 29 years	103,048,119	6,405,025 (6.2%)
25-29 years	169,767,336	10,008,136 (5.9%)
20-24 years	292,464,253	19,700,733 (6.7%)
Below 20 years	348,066,848	35,751,391 (10%)
Mother's education			<0.001
Secondary or higher	232,945,635	10,485,290 (4.5%)
Upto Primary	278,431,181	20,932,286 (7.5%)
No formal schooling	401,969,740	40,447,709 (10%)
Father's education			<0.001
Secondary or higher	278,004,208	16,678,238 (6.0%)
Upto primary	237,404,080	18,740,290 (7.9%)
No formal schooling	397,938,269	36,446,758 (9.2%)
Religion of the household			0.005
Islam	827,755,878	64,613,735 (7.8%)
Others	85,503,455	7,244,384 (8.5%)
Unknown	87,223	7,167
Wealth index quintile			<0.001
Richest	157,253,196	9,097,391 (5.8%)
Richer	177,418,951	12,390,914 (7.0%)
Middle	185,811,445	14,519,184 (7.8%)
Poorer	195,948,801	17,269,594 (8.8%)
Poorest	196,914,163	18,588,204 (9.4%)
Household structure			<0.001
Extended	430,842,462	32,421,621 (7.5%)
Nuclear	482,504,094	39,443,665 (8.2%)
Sex of household head			<0.001
Female	96,735,517	6,848,607 (7.1%)
Male	816,611,039	65,016,679 (8.0%)
Place of residence			<0.001
Urban	182,944,415	11,732,975 (6.4%)
Rural	730,402,142	60,132,311 (8.2%)
1n = Absolute distribution; % = Row percentage; Adjusted for sampling weights
2Pearson's X^2: Rao & Scott adjustment

Table 3 shows that among children whose preceding sibling died before their conception, 15% experienced infant mortality. Mortality was also higher for children whose previous sibling died after their conception. Infant mortality varied significantly across birth intervals, birth cohorts, and parental characteristics. Mortality was highest among children born within 18 months of a previous birth, with 21% experiencing infant mortality. In contrast, children born after between 25-29 years had the lowest mortality rate of 6.8%. Sex differences were minor, with female infant mortality at 7.7% and male mortality at 8.7%.

Multivariable analysis

Regression of infant deaths by background characteristics

Here we will check the multivariable association of infant mortality with the independent variables by estimating logistic regression models. We run the weighted regression models by:

• Model 1: Without main independent variables.
• Model 2: With main independent variables.

# We merge the model-1 and model-2 tables
t3_log <- tbl_merge(
    tbls = list(t3_log_mdl1, t3_log_mdl2),
    tab_spanner = c("**Model-1**", "**Model-2**")
) |> 
    # ensure that estimates of sibsurv_3c and binterval_3c_opp are at the top
    modify_table_body(
        ~ .x |> 
            arrange(desc(variable %in% c("sibsurv_3c", "binterval_3c_opp")))
    )
# Now we view the table
t3_log |> 
  as_flex_table() |> 
  fontsize(size = 9, part = "all") |> 
  theme_vanilla()

Table 4: Multivariable association of infant mortality and the independent variables among children in Bangladesh 1970-2022

	Model-1				Model-2
Characteristic	OR1	95% CI1	p-value	GVIF1	OR1	95% CI1	p-value	GVIF1
Death scarring								1.2
Alive					—	—
Dead after conception					1.32	1.21, 1.44	<0.001
Dead before conception					1.91	1.81, 2.01	<0.001
Preceding birth interval (in months)								1.2
24 and more months					—	—
18-23 months					1.58	1.50, 1.67	<0.001
Less than 18 months					2.30	2.18, 2.42	<0.001
Birth cohort of index child				1.6				1.6
2010 and above	—	—			—	—
2000-2009	1.31	1.22, 1.40	<0.001		1.17	1.07, 1.29	<0.001
1990-1999	1.96	1.84, 2.09	<0.001		1.70	1.56, 1.85	<0.001
1980-1989	2.80	2.62, 2.99	<0.001		2.40	2.20, 2.62	<0.001
1970-1979	3.57	3.31, 3.85	<0.001		3.11	2.82, 3.43	<0.001
Birth order				3.3				2.8
1-2	—	—			—	—
3	0.86	0.82, 0.90	<0.001		0.95	0.90, 1.01	0.083
4	0.97	0.91, 1.03	0.3		1.00	0.93, 1.07	>0.9
5 and more	1.23	1.14, 1.32	<0.001		1.15	1.07, 1.24	<0.001
Sex of child				1.1				1.1
Female	—	—			—	—
Male	1.21	1.18, 1.25	<0.001		1.15	1.11, 1.20	<0.001
Season during birth				1.1				1.2
Summer/pre-monsoon	—	—			—	—
Monsoon	1.02	0.98, 1.06	0.4		1.07	1.01, 1.13	0.019
Winter/post-monsoon	1.01	0.97, 1.04	0.8		1.01	0.97, 1.06	0.7
Mother's age at child birth (in years)				3.4				2.9
Above 29 years	—	—			—	—
25-29 years	1.01	0.94, 1.08	0.8		0.93	0.87, 1.00	0.049
20-24 years	1.21	1.12, 1.30	<0.001		1.00	0.93, 1.08	>0.9
Below 20 years	1.77	1.63, 1.92	<0.001		1.14	1.05, 1.25	0.003
Mother's education				2.1				2.0
Secondary or higher	—	—			—	—
Upto Primary	1.25	1.18, 1.32	<0.001		1.26	1.17, 1.36	<0.001
No formal schooling	1.39	1.31, 1.48	<0.001		1.37	1.26, 1.48	<0.001
Father's education				1.6				1.8
Secondary or higher	—	—			—	—
Upto primary	1.11	1.05, 1.16	<0.001		1.06	0.99, 1.13	0.11
No formal schooling	1.15	1.09, 1.21	<0.001		1.13	1.05, 1.21	<0.001
Religion of the household				1.0				1.1
Islam	—	—			—	—
Others	1.11	1.04, 1.17	<0.001		1.10	1.02, 1.17	0.008
Wealth index quintile				1.8				1.9
Richest	—	—			—	—
Richer	1.09	1.03, 1.17	0.005		1.06	0.98, 1.14	0.14
Middle	1.17	1.10, 1.25	<0.001		1.16	1.07, 1.26	<0.001
Poorer	1.30	1.22, 1.38	<0.001		1.25	1.15, 1.35	<0.001
Poorest	1.39	1.30, 1.49	<0.001		1.33	1.22, 1.44	<0.001
Household structure				1.1				1.1
Extended	—	—			—	—
Nuclear	1.05	1.02, 1.09	0.004		1.03	0.99, 1.08	0.13
Sex of household head				1.0				1.0
Female	—	—			—	—
Male	1.05	1.00, 1.11	0.074		1.07	1.00, 1.14	0.056
Place of residence				1.2				1.2
Urban	—	—			—	—
Rural	0.98	0.94, 1.03	0.5		0.98	0.93, 1.04	0.6
No. Obs.	349,936				230,026
AIC	184,928				111,725
BIC	185,180				112,009
1OR = Odds Ratio, CI = Confidence Interval, GVIF = Generalized Variance Inflation Factor

Table 4 presents the results of the logistic regression analysis, showing key factors associated with infant mortality. Bangladeshi children whose preceding sibling had died before their conception had 1.86 times higher odds of experiencing infant mortality. Short birth intervals, particularly those under 18 months, were associated with a significantly higher risk, with mortality odds 2.51 times higher than for births with a preceding interval of 24 months or longer.

START FROM HERE

UPCOMING TASKS:

• Run survival regression models and check the result.

• Check if frailty survival regression models can be run.

• Check and apply the evalue package.

TO BE CONTINUED …