BMJ Global Health Results Sample The final sample comprised of 67 countries and 173 nationally representative surveys which included 1.6 million children under five. Table 1 and online supplementary figure S1 show the distribution of countries and surveys included, and the percentage of children under five without birth registration for the first and most recent survey. The final sample included countries from 6 of the 7 World Bank regions ranging from 3 countries in Middle East and North Africa to 32 countries in Sub-­Saharan Africa. Most countries included were low-­middle (45%) or low-­income countries (40%) and only 15% were upper-­middle income countries. Of the 173 surveys, 75 (43%) were DHS and 98 (57%) were MICS. Surveys were conducted between 1999 and 2016: 17% before 2004 (Wave 1), 33% between 2004 and 2008 (Wave 2), 28% between 2009 and 2012 (Wave 3), and the remaining 21% were conducted between 2013 and 2016 (Wave 4). Cross-sectional analyses of children without birth registration Table 1 shows the percentage of children without birth registration based on caregiver report at the first and most recent survey. At the first survey, the smallest estimates of non-­registration were similar across regions, and in five out of six regions were below 5%, which we defined as complete birth registration coverage. However, in South Asia, the smallest estimate was 58.9% (95% CI 57.8 to 59.9) in India in 2005. In contrast, the largest estimates of non-­registration at the first survey varied widely by region: in Europe and Central Asia, the largest estimate of non-­registration was 25.4% (95% CI 22.0 to 29.2) in Tajikistan in 2000; in South Asia it was 90.2% (95% CI 89.4 to 90.9) in Bangladesh in 2006 and in Sub-­Saharan African non-­registration was highest in Liberia in 2007 where 96.4% (95% CI 95.4 to 97.2) of children under five were unregistered. Estimates of non-­ registration from the most recent surveys also varied by region and were overall lower than estimates from the first survey. However, at the most recent survey, the percentage of children without birth registration was greater than 50% in 16 out of 67 countries. Online supplementary table S2 shows estimates and CIs for each survey by country. 4 Figure 1 shows wealth, urban rural and gender inequalities in birth registration for each survey. Large wealth and urban/rural inequalities were observed in surveys in every region except Europe and Central Asia and were largest in South Asia and Sub-­Saharan Africa. The SII was significantly different from the null value of 0 representing no inequality in 129 out of 173 surveys which indicated that fewer children living in the poorest households compared with the wealthiest had their births registered. Among these surveys, wealth gaps were as large as 20 percentage points in in 89 surveys and larger than 50 percentage points in 26 surveys. The SII was greater than 70 percentage points in two countries: Nigeria in 2011 and Pakistan in 2012. Urban/rural inequalities were not insubstantial: 105 out of 173 surveys had significant urban/rural inequalities. Non-­registration was 20 percentage points or higher in rural areas compared with urban areas in 45 surveys. Niger in 2006 and 2010 was the only country with an absolute difference larger than 40 percentage points between urban and rural areas. In contrast, the percentage of children without birth registration was statistically significantly higher in urban areas compared with rural areas in six surveys. Online supplementary tables S3 and S4 show estimates and CIs wealth and urban/rural inequalities for each survey by country. Gender inequalities in registration were observed in fewer surveys and were smaller in magnitude. In 17 surveys, the absolute difference in non-­registration was statistically significantly higher among girls compared with boys, however the largest inequality was 3.9 percentage points in Mali in 2006. In five surveys, the reverse was observed, and non-­registration was significantly higher among boys—the largest gap was 5.3 percentage points in Swaziland in 2014. Changes in the percentage of children without birth registration (1999-2016) Figure 2 shows changes in the percentage of children without birth registration for each country. A total of 14 countries, of which 9 were in Europe and Central Asia, had achieved complete birth registration coverage by the first survey and sustained this until the most recent survey. Among the 53 countries that had not achieved complete coverage, non-­registration decreased in 39 countries and increased in 14 countries. Among the 32 countries in Sub-­ Saharan Africa included in the sample, non-­registration increased in 10 and decreased in 22 countries. Annual decreases in non-­registration ranged from 5.7 percentage points in Mozambique to 0.001 percentage points in Honduras. A total of 26 countries achieved an annual decrease greater than 1 percentage point, and 8 of these countries achieved decreases greater than 3 percentage points. Annual increases in non-­registration were smaller compared with decreases and ranged from 0.2 percentage point in Haiti to 3 percentage points in Zimbabwe. Annual increases in non-­registration greater than 1 percentage point were observed in four countries. Bhatia A, et al. BMJ Global Health 2019;4:e001926. doi:10.1136/bmjgh-2019-001926 BMJ Global Health: first published as 10.1136/bmjgh-2019-001926 on 16 December 2019. Downloaded from https://gh.bmj.com on 21 July 2025 by guest. Protected by copyright, including for uses related to text and data mining, AI training, and similar technologies. Changes in gender, urban/rural and wealth inequalities in birth registration To examine changes in inequalities in non-­registration over time, we estimated the absolute change in gender, urban/rural and wealth inequalities between the first and last survey, annually and between survey waves. Differences in estimates of inequality were deemed significant if the CIs between the first and most recent survey were not overlapping. All estimates were weighted to account for the multistage sampling design. Analyses were conducted in Stata 15.