Table 1.

Excluded studies and reasons for exclusion

Reference	Reason for exclusion	n (%)
Arpino 2014, Barbosa 2002, Blum 2010, Dagne 2015, DiRienzo 2009, Guan 2017, Huang 2012, Kenward 2001, Nyirenda 2010, Obare 2010, Patrician 2002, Scharfstein 2003, Sun 2018, Tian 2007 [12–24].	Not a survey	14 (31.1)
Bärnighausen 2012, Grassly 2004, Hlalele 2008, Kranzer 2008, Liu Y 2015, Liu S 2015, Mistry 2008, Nelwamondo 2007, Pantanowitz 2009a, Pantanowitz 2009b, Rosinska 2013, Schomaker 2018, Shah 2014, Westreich 2012, Wirth 2010, Wu 2001 [25–39].	Do not measure HIV prevalence	16 (35.6)
Boerma 2003, Brookmeyer 2010, Clark 2012; Garcia-Calleja 2006, Gouws 2008, Hund 2013, Korenromp 2013, Larmarange 2014 [2, 6, 40–44].	Methodological article	8 (17.8)
Alkema 2008, Montana 2008, Kayibanda 2011 [45–47].	No missing data methods used in the analysis	3 (6.7)
McGovern 2015a, Obare 2010, Pentanowitz 2009a [23, 33, 48].	Duplicate	3 (6.7)
Ng 2013 [49].	Could not assess the risk of bias	1 (2.2)