1 Monash University, Australia.
6 Like similar questions in the DHS, these questions would only identify pregnancy-related rather than maternal deaths according to the ICD-10 classification: the death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes. (WHO 2012, p.9).
7 Regions defined as Sumatra, Java-Bali, Kalimantan, Sulawesi and Eastern Indonesia. 8 Being a first birth is also often cited as a risk factor. We are unable to control for whether a birth is a first birth but some of our controls will act as proxies for this effect number of young children in the household and age of mother. Note also that Arifeen et al. (2014) in a rigorous quantitative study of maternal mortality in Bangladesh find that the risk associated with a first birth is less than for subsequent births.
9 The census questionnaire identifies an individual as working or employed if they were engaged in a job or business for a wage, salary or family gain for one hour or more in the past week.
10 Jakarta population: 10,135,030; Surabaya, East Java: 2,843,144; Bandung, West Java: 2,575,478; Bekasi, West Java: 2,510,951; Medan, North Sumatra: 2,185,789; Semarang, Central Java: 2,067,254; Tangerang, Banten: 2,001,925; Depok, West Java: 1,869,681; Palembang, South Sumatra: 1,561,959; South Tangerang, Banten: 1,436,187. 11 We define health centres to include private clinics, sub-district health centres (Puskesmas), doctor or midwife practices, village health posts (Poskesdes) and village maternity posts (Pondok bersalin desa, Polindes). Distance to health centre is defined as 0 where a health centre is located in the village.
12 There are no Poskesdes in Jakarta. 13 Years of education is estimated as follows: 0 for never attended school; 3 for incomplete primary school; 6 for complete primary school; 9 for junior high school; 12 for senior high school; 14 for tertiary education.
Aceh North Sumatra Riau Island East Kalimantan North Sulawesi Riau West Sumatra Jambi Bangka Belitung West Kalimantan Central Kalimantan West Sulawesi Gorontalo Central Sulawesi North Maluku West Papua South Sumatra South Kalimantan Bengkulu Papua Maternal Mortality Ratio (281,371] (4) (206,281] (4) (171,206] (4) (144,171] (4) (128,144] (4) (115,128] (4) (97,115] (4) [47,97] (5) Lampung Banten Jakarta West Java Central Java Yogyakarta East Java Bali South Sulawesi Sulawesi Tenggara West Nusa Tenggara East Nusa Tenggara Maluku
Aceh North Sumatra Riau Island East Kalimantan North Sulawesi Riau West Sumatra Jambi Bangka Belitung West Kalimantan Central Kalimantan West Sulawesi Gorontalo Central Sulawesi North Maluku West Papua South Sumatra South Kalimantan Bengkulu Papua Number of maternal deaths (495,1208] (4) (237,495] (4) (173,237] (4) (150,173] (4) (121,150] (4) (77,121] (4) (62,77] (4) [41,62] (5) Lampung Banten Jakarta West Java Central Java Yogyakarta East Java Bali South Sulawesi Sulawesi Tenggara West Nusa Tenggara East Nusa Tenggara Maluku
Aceh North Sumatra Riau Island East Kalimantan North Sulawesi Riau West Sumatra Jambi Bangka Belitung West Kalimantan Central Kalimantan West Sulawesi Gorontalo Central Sulawesi North Maluku West Papua South Sumatra South Kalimantan Bengkulu Papua Lampung Jakarta South Sulawesi Sulawesi Tenggara Maluku Average distance to the nearest hospital (Km.) (39.1,47.4] (5) (27.5,39.1] (6) (20.9,27.5] (5) (17.1,20.9] (6) (9.0,17.1] (5) [1.1,9.0] (6) Banten West Java Central Java Yogyakarta East Java Bali West Nusa Tenggara East Nusa Tenggara
Aceh North Sumatra Riau Island East Kalimantan North Sulawesi Riau West Sumatra Jambi Bangka Belitung West Kalimantan Central Kalimantan West Sulawesi Gorontalo Central Sulawesi North Maluku West Papua South Sumatra South Kalimantan Bengkulu Papua Average number of midwives (0.63,0.88] (5) (0.48,0.63] (6) (0.40,0.48] (5) (0.25,0.40] (6) (0.13,0.25] (5) [0.00,0.13] (6) Lampung Banten Jakarta West Java Central Java Yogyakarta East Java Bali South Sulawesi Sulawesi Tenggara West Nusa Tenggara East Nusa Tenggara Maluku
Aceh North Sumatra Riau Island East Kalimantan North Sulawesi Riau West Sumatra Jambi Bangka Belitung West Kalimantan Central Kalimantan West Sulawesi Gorontalo Central Sulawesi North Maluku West Papua South Sumatra South Kalimantan Bengkulu Papua Lampung Jakarta South Sulawesi Sulawesi Tenggara Maluku Years of education of household head (9.3,10.7] (5) (8.8,9.3] (6) (8.3,8.8] (5) (8.0,8.3] (6) (7.6,8.0] (5) [6.7,7.6] (6) Banten West Java Central Java Yogyakarta East Java Bali West Nusa Tenggara East Nusa Tenggara
k y i = β0 + β i x i + ε i, y = 1 [y > 0], i=1 y i i x i ε i 14 A logit model is an alternative method to estimate the effect of covariates on indicator variables. We choose a probit model as the marginal effects are easier to interpret as percentages. 15 Recall that the census birth data was collected on ever-married women, while the death data was collected on all women. This complicates the interpretation somewhat as to whether we refer to all at-risk women or only ever-married at-risk women.
16 The fact that we cannot include at risk women whose pregnancy did not progress to a live birth could bias our results if the relationship between the household and village characteristics and maternal mortality differ for these women (compared to women whose pregnancy did produce a live birth).
17 The decomposition is conducted using the Oaxaca command in STATA with the probit option. The decomposition uses the results of the reported probit estimation in Table 5 to predict what maternal mortality ratios would be in the low performing regions if they had the characteristics of the better performing regions. It does this by substituting the means of the variables in the better performing regions in to the estimated relationship between the variables and maternal mortality for the poorer performing regions. 18 Table A4 in the appendix presents comparable results for the high-performing provinces. In general the health access variables are found to be less protective. This is likely due to the better and less variable access in these provinces.
MAMPU (Empowering Indonesian Women for Poverty Reduction) is a joint initiative between the Indonesian and Australian governments.