Table 1 Description and Measurement of Exposure Variables

Optimum maternal healthcare service utilisation

The principal independent variable was ‘maternal healthcare service utilisation’. It was generated using the variables ‘antenatal care visit’, ‘tetanus vaccination’, ‘place of delivery’, ‘skilled birth attendance’ and ‘postnatal check-up’. Scores were assigned to the responses of each woman to each of the questions: number of antenatal care visits (none = 0; 1–3 = 1; ≥4 = 2), skilled delivery (no = 0; yes = 1), postnatal check (no = 0; yes = 1), received the required number of tetanus injections during pregnancy (no = 0; yes = 1), place of delivery (home = 0; other facilities = 1; modern health facility = 2). This produced a maximum overall score of 7 and a minimum of 0. Following this, the overall total score for eligible respondents was disaggregated by quartile. 2016 Ethiopian Demographic and Health Survey Data were collected health-related information five years before the survey that means all the data were collected before WHO introduced the new recommended antenatal visits.

\( \kern1.25em MCHI=\kern2.5em \left\{\begin{array}{c} None, if\kern27.75em x=0\kern1.25em \\ {} Low, if\kern8.5em 0<x<50\% of\ the\ \mathit{\max}. overall\ score\kern5em \\ {} Medium, if\kern9.25em 50\%\le x<75\% of\ the\ \mathit{\max}. overall\ score\\ {} High, if\kern10em 75\%\le x<100\% of\ the\ \mathit{\max}. overall\ score\ \end{array}\kern1em \right. \)

Infant sex

Categorised as ‘male’ or ‘female’.

Preceding birth interval

This is the difference in months between the current birth and the previous birth, counting twins as one birth; it was based on the date of birth of the children based on the mother’s self-reporting [36, 37]. It was re-coded as (i) ‘less than 24 months (or short birth interval)’, (ii) ‘between 24 and 59 months (or recommended birth interval)’ and (iii) ‘greater than 60 months’ [36, 37].

Reported birth size

Mothers were asked to estimate the size of their child based on their experience or by comparison with a previous child. The response options were ‘greater than average’, ‘average’, ‘smaller than average’ and ‘very small’. This variable was re-coded as ‘above average’, ‘average’ and ‘below average’ (smaller than average or very small) [36, 37].

Multiple births

This is the number of births for one pregnancy, and the response option was open to any number of births. This was re-coded as ‘single birth’ for one birth or ‘multiple birth’ for more than one birth [36, 37].

Total live births

Participants were asked how many live births they had had in their lives, and the response option was continuous. This was re-coded based on quartiles as (i) ‘one live birth’, (ii) ‘two to three live births’, (iii) ‘four to five live births’ and (iv) ‘more than five live births’. If the respondent was pregnant during the interview, one was added to the total live births [36, 37].

Maternal age

Women were asked their age (in years) when completing the survey. It was then categorised as ‘15–19’, ‘20–24’, ‘25–29’, ‘30–34’, ‘35–39’, ‘40–44’ and ‘45+’. F For this study, the reproductive age of the mother was re-coded as ‘young’ (15–24 years), ‘young adult’ (25–34 years) and ‘middle aged’ (35–49 years) to obtain adequate samples for each category [36, 37].

Religion

Participants were asked to report their religion as either ‘Orthodox’, ‘Catholic’, ‘Protestant’, ‘Muslim’ or ‘Traditional or other’. Orthodox, Catholic and Protestant were re-categorised as ‘Christian’; the categories of ‘Muslim’ and ‘Traditional or other’ were retained [36, 37].

Maternal education

The original question was ‘what is your highest level of education?’, and the response options were ‘no education’, ‘primary education’, ‘secondary education’ and ‘higher education’. We re-coded these as ‘no education’, ‘primary education’ and ‘secondary education and above’. We merged secondary and higher education because of the small numbers in the higher education category [36, 37].

Decision-making autonomy

The original questions were ‘who usually decides to obtain healthcare?’, ‘who usually decides on the purchase of household items?’ and ‘who usually decides to visit relatives?’. The response options were ‘respondent’ or ‘partner’. A new variable was created as a combination of the three responses. The value of each characteristic was either 0 (partner) or 1 (respondent). If the respondent scored 3 out of 3, they were categorised as ‘yes’ to having decision-making autonomy and ‘no’ if otherwise [36, 37].

Media exposure

The original questions for this variable were ‘do you listen to the radio at least once a week, less than once a week or not at all?’, ‘do you read a newspaper/magazine at least once a week, less than once a week or not at all?’ and ‘do you watch television at least once a week, less than once a week or not at all?’ [37]. A composite variable was created combining whether a respondent read newspapers/magazines, listened to the radio and/or watched TV. This was categorised as ‘no access’ if the woman had no access to any of the three media, ‘1’ if the woman had access to any of the three media less than once a week, and ‘2’ if the woman had access to any of the three media at least once a week [36, 37].

Number of pregnancy losses (miscarriage, abortion or stillbirth) and child deaths

Participants were asked if they ever had a pregnancy that miscarried, was aborted or ended in a stillbirth, and if they had ever successfully given birth to a child who later died. A new variable was created as a combination of child mortality (from the birth history question) and pregnancy loss (from the pregnancy history question). To generate this new variable, the number of child deaths from the birth history was first generated. Second, the number of pregnancy losses (miscarriage, abortion or stillbirth) was generated. The status of the index child was excluded from the calculation. Finally, the above two values were added to create the new variable with the following categories: ‘no adverse pregnancy events’, ‘one adverse pregnancy event’, ‘two adverse pregnancy events’ and ‘three or more adverse pregnancy events’ [36, 37].

Sex of the household head

Categorised as ‘male’ or ‘female’.

Household wealth index

In the EDHS, wealth index was calculated based on household assets, such as televisions and bicycles. Principal components analysis was applied to generate the wealth index as a continuous scale of relative wealth. Wealth index was categorised into five wealth quintiles: ‘very poor’, ‘poor’, ‘middle’, ‘rich’ and ‘very rich’. For this analysis, we re-coded the wealth index as three categories for adequate sampling in each category: as ‘poor’ (poor and very poor), ‘middle’ and ‘rich’ (rich and very rich) [36, 37].

Distance to health facilities

Distance to health facilities was based on participants’ subjective ratings. The original response options were ‘distance to health facilities is a big problem’ or ‘distance to health facilities is not a big problem’ [36, 37].

Place of residence

This was originally recorded as ‘rural’ or ‘urban’ and not changed for this analysis [36, 37].

Region of residence

Defined as the region in which the infant’s mother was raised. The variable was re-coded as ‘agrarian’ (encompassing Tigray, Amhara, Oromia, Benishangul, SNNPR, Gambela and Harari), ‘pastoralist’ (Afar and Somali) or ‘city dweller’ (Addis Ababa and Dire Dawa). An agrarian society is any community whose economy is based on producing and maintaining crops and farmland. A pastoralist society is any community whose economy is based on raising livestock. A city-dweller society is any city community [36, 37].

Maternal employment

Women were considered to be employed if they had done any work other than housework in the 12 months prior to the survey and if they were paid for their labour in cash or in kind. The response options were ‘employed for cash’, ‘employed not for cash’ and ‘unemployed’. This was re-categorised as ‘employed’ and ‘not employed’.