Were the Exemplars top performers in socioeconomic equity?

A survey by my organization, the International Center for Equity in Health (ICEH), found that the following countries achieved the most marked reductions in socioeconomic inequalities in under-five mortality: Democratic Republic of Congo, Sierra Leone, Mozambique, Kyrgyzstan, Comoros, Kenya and Malawi.

To access the full study, please click here.

The study relied on approximately 400 survey datasets from either Demographic and Health Surveys (DHS) or Multiple Indicator Cluster Survey (MICS). For more information on the methodology employed by these surveys: DHS and MICS. Both types of survey programs are highly comparable in terms of sampling and questionnaires. A few national surveys that are not standard DHS or MICS were included when their data allowed estimates of under-five mortality.

Indicators

U5MR (the under-five mortality rate) was calculated as follows. Women aged 15 to 49 years in the survey samples were asked about pregnancies and deliveries in the years preceding the survey, including characteristics such as birth date, sex of the children, and survival status. If a child was not alive, the age at death was recorded. U5MR estimates were based on the ten years preceding the survey. The ten-year period is required for calculating death rates by wealth quintile (see below) with sufficient precision. Mortality rates are presented as the number of deaths per 1,000 live births. For a survey that took place in 2018, for example, the mid-point of the U5MR estimate was five years earlier, or 2013. This is important to bear in mind when interpreting the results, particularly when comparing them with estimates of current mortality levels.

The socioeconomic position (SEP) of households is based on asset indices, obtained from information on household appliances, characteristics of the building materials, presence of electricity, water supply and sanitary facilities, among other variables. Because relevant assets and their importance may vary in urban and rural households, separate principal component analyses are carried out in each area, which are later combined into a single score using a scaling procedure based on linear regression where the dependent variable is a score obtained for all households, excluding items not applicable to either urban or rural areas. The resulting score, which is comparable between urban and rural households, is then divided into quintiles.

The slope index of inequality (SII) is a measure of absolute inequality, derived from mortality rates according to family wealth quintiles. A value of -10, for example, indicates that U5MR is estimated to be 10 deaths per thousand lower at the top of the socioeconomic scale than at its bottom.
For each country, we used data from all available surveys to estimate the annual change in the SII using weighted least square regression with sample sizes as the weights. Annual change was then divided by the value of the SII in the earliest survey for each country and expressed as a percentage.

Results

A total of 58 countries had more than one survey over time for which it was possible to estimate U5MR by quintile. Of these:

  • 10 countries were excluded because the most recent survey was in 2011 or earlier.
  • Ethiopia was excluded because the SII in the earliest survey was slightly positive (higher mortality among children from wealthy families) and became slightly negative over time, indicating that inequality increased.
  • 7 countries (Albania, Burundi, Chad, Liberia, Tajikistan, Togo and Zimbabwe), had negative SII at baseline which increased over time, also indicating that inequalities worsened.
  • 10 countries (Armenia, Colombia, Gabon, Honduras, Jordan, Lesotho, Maldives, Namibia, Tanzania and East Timor) presented SII values that were very small – less than 5 deaths per 1,000 – at baseline, and were also excluded.

The table below shows the remaining 30 countries, ranked according to the percent of the equity gap in U5MR closed.

The top performers were DR Congo, Sierra_Leone, Mozambique, Kyrgyzstan, Comoros, Kenya and Malawi.

Limitations

When interpreting these results, the following limitations must be considered.

  • Data were only available for 58 countries.
  • The number of surveys per country varied, as did the time elapsed from the first to the latest surveys. Countries with a larger number of surveys spread over a longer time period allow a more precise estimate of time trends than countries with fewer surveys over a shorter time.
  • Mortality rates, as mentioned above, refer to the 10-year period prior to each survey, therefore there is a lack of recent data on inequality. Countries that made recent progress, e.g. after 2010, may not be detected among the top performers.
  • Like all statistical estimates, U5MR and SII are affected by sampling error; the Excel table provides information on the standard errors of all estimates

Country

Earliest survey

Latest survey

Years elapsed

Annual reduction in SII

% gap closed by year

Year

U5MR

SII

Year

U5MR

SII

Congo_Democratic_Republic

2007

155

-9.0

2013

111

-3.6

6

0.91

10.0%

Sierra_Leone

2008

168

-6.5

2013

175

-3.9

5

0.53

8.1%

Mozambique

1997

218

-16.5

2015

41

1.4

18

1.01

6.1%

Kyrgyzstan

1997

76

-5.4

2012

32

-1.1

15

0.29

5.3%

Comoros

1996

113

-6.8

2012

50

-1.2

16

0.35

5.2%

Kenya

1993

93

-9.9

2014

55

-1.2

21

0.46

4.7%

Malawi

2000

202

-8.2

2015

74

-2.7

15

0.37

4.5%

Niger

1998

302

-9.9

2012

152

-2.7

14

0.38

3.9%

Benin

1996

182

-9.2

2017

102

-5.4

21

0.35

3.7%

Uganda

1995

156

-7.5

2016

72

-4.1

21

0.28

3.7%

Mali

1995

252

-14.3

2012

104

-5.9

17

0.48

3.4%

Dominican_Republic

1996

61

-8.1

2013

34

-3.4

17

0.27

3.3%

Bangladesh

1993

148

-10.7

2014

54

-2.8

21

0.34

3.2%

South_Africa

1998

57

-8.3

2016

50

-3.7

18

0.26

3.1%

Indonesia

1997

71

-9.5

2017

34

-3.0

20

0.29

3.0%

India

1998

101

-11.8

2015

52

-6.4

17

0.31

2.7%

Haiti

1994

140

-6.2

2016

82

-2.8

22

0.16

2.6%

Egypt

1995

95

-14.3

2014

30

-2.6

19

0.34

2.4%

Peru

1996

68

-11.1

2018

18

-1.7

22

0.26

2.3%

Zambia

1996

193

-8.8

2013

80

-4.5

17

0.20

2.2%

Pakistan

2006

93

-7.2

2017

78

-5.6

11

0.15

2.1%

Nepal

1996

138

-8.5

2016

46

-4.4

20

0.18

2.1%

Senegal

1997

139

-14.4

2017

59

-4.9

20

0.30

2.1%

Philippines

1993

63

-8.4

2017

28

-4.0

24

0.17

2.0%

Cambodia

2000

121

-9.7

2014

47

-7.4

14

0.16

1.6%

Nigeria

2003

217

-21.0

2013

143

-15.7

10

0.34

1.6%

Ghana

1993

133

-11.1

2014

70

-3.9

21

0.17

1.5%

Guinea

1999

194

-12.7

2012

132

-10.4

13

0.18

1.4%

Rwanda

2000

208

-7.6

2014

65

-5.4

14

0.10

1.3%

Guatemala

1995

79

-6.0

2014

38

-4.5

19

0.02

0.4%

Authored by: Dr. Cesar Victora, Emeritus Professor of Epidemiology, Federal University of Pelotas, Link to Bio