BMJ Global Health There are a number of limitations to this study. Although we rely on the most recent data available, the most recent surveys in our sample are from 2015 and may not reflect the most current coverage of birth registration. Notably, we do not examine access to birth certificates—documentary proof of registration—which, prior research indicates is likely to be lower than access to registration in many countries.10 44 The sample of countries included reflects (1) the focus of the DHS and MICS programme on LMICs, (2) when questions on birth registration were included and removed from these surveys especially since birth registration questions from national surveys are often removed as CRVS systems improve, and (3) which countries had more than one survey available. In response to these limitations, we refrained from pooling data to construct regional or global estimates, and we estimated annual change to make countries comparable. Given that the annual change approach treats all annual changes as the same regardless of when they occurred, we also estimate change between survey waves to partially address this limitation. The DHS and MICS sampling approach and data structure could also affect our results. First, there could be some measurement error as caregivers are asked whether a child’s birth is registered with the ‘civil authorities’, a term which could differ in its interpretation across countries and across survey years, especially if the physical birth certificate is not available. However, this should not affect estimates within countries. Second, the MICS and DHS data used do not include children living outside Bhatia A, et al. BMJ Global Health 2019;4:e001926. doi:10.1136/bmjgh-2019-001926 households who are more likely to be marginalised and vulnerable and also do not include information on infant deaths, especially neonatal deaths, for whom birth registration is often overlooked.6 44 It is therefore likely that we underestimate the percentage of children living without birth registration and the invisibilities which drive this under-­report will persist as long as household-­ based national surveys are the primary—and only—data source in LMICs to investigate birth registration. Third, wealth quintiles represent relative rather than absolute wealth, so the poorest quintile in one country could be wealthier than the second or third quintile in another country. Other data limitations that affected our analytic approach include the lack of data on child’s age at registration, making it challenging to estimate which children were registered in a given year and to estimate yearly trends in registration. There is also no information on birth registration among children above 5 years of age, on location of registration and on facility birth for children who are not living with their biological mothers: this limits the development of detailed subnational recommendations on where birth registration access could be improved. We also recognise that a registered birth and a functioning CRVS system may not translate to the publication and use of vital statistics. However, complete registration does represent an important step. Finally, there are limitations in the choice of measure to estimate inequalities: absolute differences are sensitive to changes in the number of individuals in each stratification category and to effect modification by 9 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. Figure 3 Change in (A) wealth and (B) urban/rural inequalities among countries which increased and decreased non-­ registration. Figure shows the number of countries which achieved changes in (a) wealth inequalities and (b) urban/rural inequalities relative to changes in non-­registration.Notes: Supplementary Tables S3 and S4 show the magnitude and direction of total change in (a) SII and (b) urban/rural inequalities, annual change, and changes between survey wave in all 67 countries.