Statistical analysis of socioeconomic and demographic correlates of perinatal mortality in Tigray region, Ethiopia: a cross sectional study

This study was based on secondary analysis of retrospective cross-sectional study based on data from 2016 Ethiopian Demographic and Health Survey. The survey was carried out by the Central Statistics Agency (CSA), the Ministry of Health (MOH) and the Ethiopian Public Health Institute from January 18, 2016 - June 27, 2016. It was funded by the United States Agency for International Development (USAID). The data were obtained from the DHS measure Program with permission [9]. The main aim of this survey is to deliver detailed information on fertility, family planning, infant, child, adult and maternal mortality, maternal and child health, nutrition and knowledge of HIV/AIDS and other sexually transmitted infections to policymakers and planners.

The survey had information on a range of socioeconomic and demographic factors of the population nationwide. It implemented a two-stage sample design. The country has nine regions and two administrative cities. In the first stage 645 enumeration areas (202 in urban areas and 443 in rural areas) were selected with probability proportional to size. Second, stage involved selection of 28 households per cluster of equal probability systematic selection of the newly formed household list. The EDHS 2016 has three questionnaires: the household questionnaire, the woman’s questionnaire and the man’s questionnaire. All women of childbearing age (15–49 years) and who were either stable residents of the chosen households or visitors, who lived in the household at least one night before the survey, were eligible for the interview. Data were collected by face-to-face interviewing women that met the eligibility criteria.

A total of 15,683 women of the reproductive age group, 12,688 men aged 15–59 years in 16,650 households were interviewed in 2016 EDHS from these, 1682 households were from Tigray region. In the region 4428 births were reported in the last five years preceding the survey. Next, child with information on early neonatal death and stillbirth during the last five years were identified [10]. Finally, a total of 2738 children that have completed information about all the factors considered were included.

Tigray region is located between 36 and 40 degrees east longitude north of Ethiopia bounded in the north with Eritrea, in the east Afar region, the south Amhara region and the Sudan in the west. The 2007 Ethiopian population and housing censuses reported that the Tigray region had a population size of 3,136,267 of which 1,594,102 was females. About 85% (2,667,789) of the population are rural residents [11].

Determinants of perinatal deaths in this study were selected from the available prior studies on the subject. Important mother and child demographic characteristics explored from a theoretical perspective and various literatures for perinatal survival are presented in Table 1 [12‐15]. Wealth index was coded as: 1 = poor (includes poorest and poor), 2 = middle and 3 = rich (include richer and richest). The two categories have been merged since the percentage distribution of events (perinatal death) in the poorest and richest categories is too small. Similarly, the percentage of women in the age category 15–19 and 20–24 after screening for missing variables is too small and authors merge these two categories with age group of 25–29 years old. Since this study included only mothers who gave birth in the past five years preceding the survey, the percentage of mothers who gave birth between ages 15–24 are too small. This too few frequencies may, in turn, affect the parameter estimation. Table 1.

During the conduct of the DHS, mothers were asked to report any pregnancy lost and death of a live born infant. Then, the perinatal mortality will be obtained as the sum of the death of infants within the first week of life and pregnancy loss occurring after seven completed months of gestation. The dependent variable of this study was perinatal death (stillbirth and early neonatal mortality). Thus, the outcome variable for the i^th child is dichotomous, represented by a random variable Y_i that takes the value “1” with probability of success (perinatal death) and the value “0” with probability of failure (no perinatal death), such that

The study used descriptive statistics such as frequency distributions and percentages to describe the sample information. Multivariate logistic regression was used to estimate the effect of socioeconomic and demographic factors on perinatal death (odds ratios with their 95% confidence intervals). The goodness of fit test was done using the likelihood ratio test (LRT) and Hosmer_lemeshow tests. The degree of collinearity was checked using the variance inflation factor (VIF).

To identify the risk factors associated with perinatal mortality, 2738 births within the five years before the survey was considered. The result from Table 2 showed that out of 2738 births included in the study, 170 (6.2%) were reported perinatal deaths, of which 111 (4.1%) were early neonatal deaths and 2.1% were the stillbirths. This shows that the ratio of early neonatal deaths to stillbirths is 2/3 to 1/3. The possible explanation for this variation might be that stillbirths are likely being missed as women are afraid of reporting pregnancy terminations (fetal deaths) due to some cultures and traditions in the community. Similarly, from the total population of perinatal deaths 65.3% were early neonatal deaths (deaths before one week of a life after birth) and 34.7% were stillbirths (fetal death after 28 weeks of gestational periods). Therefore, the analysis finds that the prevalence of perinatal mortality rates was 62 per 1000 births for the five years preceding the survey.

Table 2 presented the percentage distribution of the selected risk factors of perinatal mortality in the study. About one-fourth of mothers’ were in the age group 35–39 (27.6%), and nearly one-tenth of mothers’ were below or equal to 29 years, 18 and 18.3% were in other age groups 30–34 and 45–49, respectively. More than three-fourth (76%) of mothers were illiterate (have no education at all), and only one-fifth (20%) and 4% of the mothers were attained a primary and secondary or higher education. Regarding partner’s or husband’s education level, more than half (56%) were illiterate, 37 and 6.9% of husbands attained a primary and secondary or higher education, respectively. On the other hand, when the sex of household head was concerned, 94% were reported male household head.

About one-third (33.4%) of the births in the study had belonged to a woman with poor wealth index, one-fourth (25.6%) were belonged to rich wealth index and 41% were in the middle wealth index group. Concerning mother’s occupation, nearly half of the mothers were working in the non-agricultural sector, 36.2% did not work in any sector and 21% were working in the agricultural sector.

The majority of the mothers (46.5%) had 6–8 children in their lifetime, 17.9% of respondents had at least eight children and 12.5% had less than four children in their life. Approximately half of the study population had birth orders three or less (46.2%), 29.7% had 4–5 birth orders and 24% had six or more birth orders. Regarding the previous birth intervals, 5.6% had 15 or less months of birth intervals, 23.1, 36.2 and 35% had a birth interval of 16–26, 27–38 and at least 39 months, respectively. Table 2 reveals that more than three-fourth (88%) of the respondents resided in rural areas. It also indicates that nearly one-third of respondents was using the unprotected well water for drinking and more than half (54.6%) had no toilet facilities.

The highest proportion of perinatal deaths occurs in the mothers of age groups 45–49, 35–39 and 40–49, respectively. More mothers get older more they experience perinatal deaths. Mothers with no education at all experienced the highest percentage of perinatal mortality (4.5%). As far as partner educated is concerned, the proportion of perinatal death decreased with an increase in the education level of partner.

Wealth index was inversely proportional to perinatal mortality, as wealth index increases from poor to middle and then to rich, perinatal mortality decreases from 2.7 to 0.9%. Thus, those infants born to low wealth index groups have the highest percentage of perinatal mortality than any of the higher wealth index group. Children belong to a mother who work in non-agricultural sector had a relatively higher percentage (2.4%) of perinatal mortality.

The percentages of perinatal mortality were the highest among women that had at least nine children (11.4%). A child with the smallest birth order had the highest chance of being died the first week of life. Similarly, children with birth intervals less than 15 months encountered the highest percentage (77.9%) of perinatal mortality.

Mothers living in urban were educated and had a higher access to public health care services thus, they were aware of the benefit of using this health care services in keeping the health of their babies than mothers of rural areas. The percentages of perinatal deaths in rural and urban areas were about 6.5 and 3.6%, respectively. Concerning sanitation indicator variables, the proportions of perinatal mortality was higher among mothers using the unprotected drinking water than that of piped drinking water users. Similarly, perinatal mortalities were higher among mothers with no toilet facility/bush/field than that of with facilities. Table 2.

Binary logistic regression analysis was used to examine the effect of each covariate on the perinatal death.

We start here first by checking the overall goodness of fit using the LRT and Hosmer -Lemeshow test. Accordingly, the likelihood ratio test, provided a chi-square value of 406.618 (p-value < 0.0001), which would imply good fit for the model. Similarly, the Hosmer-Lemeshow test is found the observed data was better explained by the model (chi-square value = 7.577 with 8 degrees of freedom and p-values = 0.476). Further, since the VIF value for all predictors is < 10 there is no collinearity problem.

The multivariate analysis of risk factors associated with perinatal mortality was presented in Table 3. The result showed birth type, sex of a child, previous birth intervals in months, maternal age, place of residence, mother’s education level, wealth index, source of drinking water, availability of toilet facilities, sex of household head, parity and maternal occupation were statistically significant predictors of perinatal death. Whereas birth order, husband/partner education level and maternal HIV status were not statistically significant.

According to the results from Table 3, the odds of perinatal mortality were lower among children that were singletons than those in multiple births (AOR = 0.179; 95% CI: (0.1–0.321)). Compared to a female child, males were more likely to die before celebrating their first week of life keeping all other variables constant. The odds of perinatal mortality were 74% times higher among male children than the females (AOR = 1.74; 95% CI: (1.234–2.454)).

Concerning birth interval, the odds of perinatal mortality were higher among children with a previous birth interval of less than or equal to 15 months and previous birth intervals of 16–26 months (AOR = 7.902; 95% CI: (4.526–13.795)) and (AOR = 2.088; 95% CI: (1.292–3.375)), respectively, compared to that of a previous birth interval at least 39 months. There was no significant difference in the risk of perinatal death between birth intervals 27–38 months and 39 months or higher birth interval. Table 3 also shown that the odds of perinatal mortalities were lower among children born to mothers’, whose age were less than or equal to 29, 30–34, 35–39, and 40–44 (AOR = 0.293; 95% CI: (0.128–0.668)), (AOR = 0.335; 95% CI: (0.173–0.646)), (AOR = 0.381 95% CI: (0.231–0.628)) and (AOR = 0.585; 95% CI: (0.360–0.948)), respectively, compared to that of 45–49 years old.

Place of residence was found significantly associated with perinatal mortality. The odds of perinatal death were about 64% times lower among children in urban areas than that of the rural residents (AOR = 0.364; 95% CI: (0.136–0.970)). Availability of toilet facility had a negative impact on perinatal death. The odds of perinatal mortality were 65% times higher among children belonged to a mother without toilet facilities than that of those with toilet facilities (AOR = 1.649; 95% CI: (1.093–2.488)).

For the variable sex of household head, the odds of perinatal mortality were about 62% times lower to children from the male household head than the female counterpart (AOR = 0.378; 95% CI: (0.213–0.670)). As far as parity is concerned, the odds of perinatal mortality were 85% times lower among children, whose mothers had 4–5 total children ever born and 60% time lower among children whose mothers had 6–8 total children ever born compared to that of whose mothers who had at least nine children ever born (AOR = 0.148; 95% CI: (0.211–0.828)) and (AOR = 0.401; 95% CI: (0.257–0.626)) respectively.

Finally, occupation of the mother was found a significant predictor of perinatal mortality. Compared to children whose mothers were working in the agricultural sector, those children whose mothers were working in non-agricultural sectors had significantly higher odds of perinatal mortality (AOR = 2.734; 95% CI: (1.773–4.216)). Table 3

This study used a nationally representative survey dataset, which enhance inferences for the entire country level. However, this study is based on secondary data the major limitation is; the study was a mother’s recall of events that took place for the past five years preceding the survey, which is subject to recall bias. Probably this may be women are afraid of reporting pregnancy terminations (fetal deaths) due to some cultures and traditions in the community.