Study population and variables
Our study has been based on most recent Multiple Cluster Indicator Survey, 2019 data, which is nationally representative cross sectional study [19]. Information was collected from individual level (married women at their reproductive age) and community level. The survey was conducted in collaboration with Bangladesh Bureau of Statistics (BBS) and UNICEF Bangladesh from January to May 2019. A two-stage stratified cluster sampling method was used to select the sampling units and a total of 64400 households were enumerated. Among them, 3220 Primary Sampling Units (PSUs) were selected for sample survey. Data were collected from eight divisions and 64 districts in Bangladesh, on demographic or socio-economic characteristics such as marriage, fertility, maternal age, maternal education, child mortality, family planning, breastfeeding, information about HIV/AIDS, maternal health care etc.
In this study, data were collected from 23099 women at their reproductive age (15-49 years). Data on 9748 children aged below 5 years were generated from the interviewed women. Complete birth histories of the children were collected including months and years. These data were used to find out the number of children born in the last 5 years preceding the survey and child age at death (from 2014 to 2019). Among those information, 5112 children were considered for analysis due to incomplete interview or non-response (Figure 1). Those children who died before their fifth birthday were considered as death/uncensored cases and those who were still alive before their fifth birthday were considered as alive/censored cases.
The primary outcome variable of this study was child survival status classified as being alive (coded as 0) or dead (coded as 1). The primary potential modifiable risk factors were considered in this study include mothers age at first birth (< 20 years, >=20 years), mother’s education (pre-primary or no education, primary, secondary and higher), area (urban, rural), division (Barisal, Chittagong, Dhaka, Khulna, Rajshahi, Rangpur, Sylhet), sex of child (boy, girl), babies size at birth (very large/average, very small), place of delivery (home, hospital/clinic), wealth index (poor, middle, rich), received antenatal care (ANC) during pregnancy (yes, no), caesarean delivery (yes, no), birth order (1, 2-3, 4+), previous birth interval (1st birth, <2 years, <3 years, +4 years), birth status (multiple birth, single birth), source of pure drinking water (pipe drinking water, non-pipe drinking water) and type of toilet facility (flush/ improved facility, none/non-improved facility
Models
Product-Limit (P-L) method: Product-Limit method proposed by Kaplan and Meir [20] is widely used in survival analysis for estimating the survival function and can deal with censored life time data. Suppose the event of interest occurs at k distinct time points t1 < t2 < ... < tj < ... < tk. If nj and dj be the number of individuals at risk of failure and the number of individuals failed at time tj; j = 1,2 ..., k, respectively, then the Product- Limit estimate of the survival function S(t) is given by
![](data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAO8AAABQCAIAAAB+jrVOAAAABGdBTUEAALGPC/xhBQAAAAlwSFlzAAAOwwAADsMBx2+oZAAACKFJREFUeF7tm2113DAQRUMhGEohHAohGEKhDMogDIqgCEogBMKgHNq7frM6ir4se72Orcz90bMjy7I08zQaO6cP/xxnFFzNzji4mp1xcDU74+BqdsbB1eyMg6vZGQdXszMOrmZnHFzNzji4mp1xcDU74+BqdsbB1eyMg6vZGQdXszMOrmZnHFzNzji4mp1xcDU74+BqdsbB1eyMg6vZGQdXszMOrmZnHFzNzkH58ePHw8PDnz9/zO7A1ex8Gt++fUOv8P7+bk0R379/55IZfbianc/k6enp8fHRjJvZXs1vb28/f/5kltp27D+OjN+/f//9+9d6OM4VpPz8/GzGzWyjZpT669evl5cXJicRF+HseH19Re52m/P1oA5WCQHomH/JfXbtCi3qgKisqY+b1MzM4jQMTBRN08glJWN+MCdauBTqJERPN9o9YX8piDjRJ/TEHaSH4nueXgGXJr71akbEsXaLhXyRRN8UIXbBGRoUgkDjukJJ2oyPKG2b0c02lYbjzELiQ6Bx1lNCNOMjpLnapQauZmcnKC+RrxnTuxbipqIwO6Jxqc291MyEqCXYYbMVCFUHi+Rfs3dBZdkiVOvb/c5y8GGcbl9fX2kp1pk0cmmFJO6iZmajjdhZEzNvdA97fu6gfL+IdMKaSqi2g3PV95+SI9rgw6Dm8ApYTHbkQS7FYiCPUEkX3xdjetXMQAwXPsDxg5nxVLscgQfpsDSNsTy8z7B7ClprAbNLBDWbfXhwoDxJIPCqtUYgIKKDmMzeC6bEQ5kSWmICOh51hie7Tj43Y0LdpLriokRXkHg2o6NmSY3h1JI8EpTw4vfWfhiW1TLj2eJkK7QEMLvEudRMgHAgai6GXDrWcliXte4FkmVuPJo5MD2kggl5TqxNr706mA+Scm0iUIYrPrJxfAjGaXS4ZTOsgGcJs0ucSM0EhWAr/1nTlaBjfEsffuyv5k6QLNOrvQLqKoI2+yPzQZJA86qREZNdNVtj4FM6MKDZJRiWPjoE7g0PEmaXOJGapdfk1BZEiqtKIvym22HVrOk1BKAOvESaHTEfJO4EsqbZddpCxJXq0JA7zG6JDbksbMLsEmdRc0+mEEdWM+Lh6GjPTXUB3cyO6FUzLjC7grxZfAbodMvJN0l7nG3RHMDsEmdRs16qihkr4ZhqVglB3FlIrSwO1E6h+SCp0oX2Jyrl1LaPNI7Ouwbq1jhrVF730N6E1mkINasgbDgtcOTc3InElr9fzQdJyf8Sz6Y45KNGhSAJ9iRdCWiH77uXJU2YXeIUataB1jnJAdRc01LX+hG0tj5Q+xYPAiX/htz1t5+e7xUSUGOorbisZ8LsEqdQs6LbUzTDIjX3H4OC/nbnndHjEin2Bonbwl+Di4KelaAqlh6Nnk7N9Kl9Uaqx4pYGPWVe4FPUbJfXYqNEqD153LKUI69BLjVFvSFBlSs9e3d2qH4Xt7eEdbpNzZrMUmlyS6eeelgk0EWdt4In3oKNElFsX6ZmUEWR+6JdaSwq7CSgthA3QVMCs0toMo0+HFMIuvFqi9BzrXNLzxsbfXBs7G0SSu7846t5c6aY3Kzmmi/UXiuLldQpUcxuwuB0HuYtkHtXSIflh0fHt6vRjCv3qzQOy8Uv/WrWS5sZEfJF/u2i7dC21hPoCT2p60b0IDC7hNRT68NVqK2LhK23Baos9VSSjn8nkOnxfPzOjWPDJbkRNALokqqdnu9F8BXVrMohOUBxqBydH6ztWkIeDPFjHLxZPJ0RMT07A3Mj03wvmF2CeTb6kEG5FCuD+WNqpSxQOkOU/ABtUWWKRE905i69XQD+p79dm1BJw1BcnQa7oEtt5ycoFp3n5DFh4SwhF0l1/VItaw45kshhMkTIFglydLFCUPwYk5DgenoWMxOoZ5777wEPEmaXaKtZMkIfZl8dHSs1MUXSqBQOuJfRGNYuZGg+ZkQoXkHfNYhmSPy42lrPhkSSH4nVQOJTekugQhJsOFrFRk2ISvbAsA2n64mzUdkEzQfMLtFWs3JzvIGVQUMKADrMqvnygL7/bVFTM6GhvZYjYHpCmX28vSHa/Lm7WoFcQa0O6QS3cnse+ztxieSE2SXaatap3V4vHTrVjPfIOuwHay1RU7NOCUYw+/xof+Y7Df/QziGWO2pjNd8iRyZHMJjl6s2wFKYqzC7RVjNX8+otgXtzhySN5PJwdjEgv+PsHqP5mPERjdCT4E9BbaXKIHF1F9hYzdCuN2ogZWoMAlmL4uZoi4vGUWs9Kt9YaM+VmlDsU2xkSgQp1LV0yKVZizFwOw4EfljTcBAFlo9UzP7I9moG6kh8yiM7CzJiRghhNykzsfDiJfJXCiYT92E5yQuuPJskCRoRXFy/xgoLC+TGXM0BHCLVAvfGr2uakhybu4sW+jPVIQU9u7q7qBl4ntLMbNmARJjfGc9HvVknEs9rLXXDFSwzHFlJnyIET8VD3JPxL5vj+gE7dy938SA6HMql6IEdzqzwA7+ZtrZlUSGqJSAsQXLSqmtShnup+SsgqSXBwNcILsma9KExDgM3EhgzmnBXsmH0iFwEMejgUDkCXzEZFiKNah/KTA43oVfA4EZM1M+qZdZwNa9H6daMhRCqvLYZHsk35FfUWVOzMrcZ3biaV6L6YV3y4y7uTTLuV0AlREixDT+QKTrPrhhX82IICY4mDMWkMguHJqFatw3ODn6j9jXjWqrFBZighfb4TboTV/NiKOaSIngR7Xp3bNBonHHZ1WBGRCivze7G1ezsRFIls6sxw0eeGBUkyZt0D65mZyf0phGK5iBufVJUo1AhZ8YSXM3OTugzRVyhSbW0J9UXjXFB0o+r2TkWpGrUvOIVEFzNzrFYXTSDq9k5EIhYf742eyGuZucQqMBAytQYq79+upqdcXA1O+PganbGwdXsjIOr2RkHV7MzDq5mZxxczc44uJqdcXA1O6Pw799/cmd62Dn6ox4AAAAASUVORK5CYII=)
Cox Proportional Hazard (PH) Model:
In this study, the risk of death in childhood was measured in months and it was a time-to-event data. There exist distinguish possible survival model options and for this study an event history analysis procedure which was proposed by Cox [21]. It is usually used to examine the impact of various factors on the risk of death. Cox Proportional Hazard (PH) Model most commonly used for analyzing censored survival data where distribution of life time is considered as unknown or unspecified. According to Cox PH, the hazard function can be defined as: h(t|X) = h0(t)* exp (β/X) ; where h(t|X) is the hazard of child death at time t, h0(t) is the baseline hazard and β = (β1, …, βm, … ,βp )/ being the p*1 vector of regression coefficients associated with in presence of a set of covariates X = (X1, …, Xm, …, Xp)/ .
Statistical Analysis
Descriptive statistics were used to summarize the distribution of selected background characteristics of under-five children. In this study, bivariate analyses were accomplished to find out the potential determinants of under-five child mortality. The prevalence of under-five child mortality according to the selected covariates was compared using Kaplan–Meier log-rank test [22] and the test has been employed to test whether the survival probabilities in different categories of a covariate are equal or not. Then with the significant factors (at p<0.05) from bivariate level, Cox PH model was fitted to assess the all possible risk factors for under-five child mortality. The results in adjusted cases were interpreted from the hazard ratios. STATA 16 employed to analyze the data.