Table 2.
Assessment of risk of bias within studies
| Study | Number of STIs tested | Sampling site for HIV | ART adherence reported |
|---|---|---|---|
| Adolf 2012 [31] | 1 | Blood plasma | No |
| Anderson 2008 [32] | 6 | Blood and CVL | No |
| Conley 2010 [33] | 6 | Not reported | No |
| Chan 2008 [12] | 3 | Blood and semen | No |
| Cu-Uvin 2001 [26] | 3 | Blood and CVL | No |
| Graham 2011 [34] | 7 | Cervix and vagina | Yes |
| Jarzebowski 2012 [35] | 1 | Blood plasma | No |
| Kofoed 2006 [36] | 1 | Blood plasma | No |
| Madeddu 2014 [41] | 1 | Blood plasma | Yes |
| Politch 2012 [37] | 4 | Semen | No |
| Sadiq 2002 [38] | 2 | Blood and semen | No |
| Sha 2005 [39] | 8 | Blood and CVL | No |
| Sudenga 2012 [40] | 8 | Blood plasma | Yes |
| Winter 1999 [10] | 2 | Blood and semen | No |
Studies reporting a large number of STIs tested, ART adherence and measuring HIV, VL viral load in genital secretions are less likely to bias estimates of the effect of STI co-infection on HIV transmission, CVL cervicovaginal lavage