Table 4. Predictors of early viral suppression (viral load<400 copies/ml) on second-line ART in the community programme.
| Viral suppression | Univariable, N = 211 | Multivariable, N = 191 | ||||
| N/at risk (%) | RR (95% CI) | pa | aRR (95% CI) | pa | ||
| Transfers into programme on ART, N = 191 | ||||||
| Yes | 82/100 (82.0) | 1.33 (1.10-1.61) | 1.33 (1.11-1.61) | |||
| No | 56/91 (61.5) | 1 | <0.01 | 1 | <0.01 | |
| Virological suppression, first-line ART, N = 150 | ||||||
| Yes | 75/108 (69.4) | 1 | 0.56 | |||
| No | 27/42 (64.3) | 0.93 (0.71-1.20) | ||||
| Reported non-adherence, first-line ART | ||||||
| Yes | 5/8 (62.5) | 0.86 (0.50-1.49) | ||||
| No | 147/203 (72.4) | 1 | 0.60 | |||
| Reason for switch,b N = 190 | ||||||
| Treatment failure | 114/159 (71.7) | 1 | 0.77 | |||
| Other | 23/31 (74.2) | 1.03 (0.82-1.30) | ||||
| Year of switch | ||||||
| ≤2007 | 74/95 (77.9) | 1 | 0.08 | |||
| 2008 | 78/116 (67.2) | 0.86 (0.73-1.02) | ||||
| Duration of viraemia, N = 206 | ||||||
| <12 months | 72/94 (76.6) | 1.11 (0.94-1.32) | 1.22 (1.03-1.44) | |||
| ≥12 months | 77/112 (68.7) | 1 | 0.21 | 1 | 0.02 | |
| VL(log10) at switch | ||||||
| ≥5 | 20/29 (69.0) | 1 | 0.59 | |||
| 4-4.99 | 80/114 (70.2) | 1.02 (0.77-1.34) | ||||
| <4 | 52/68 (76.5) | 1.11 (0.84-1.46) | ||||
| CD4 at switch | ||||||
| <100 | 32/54 (59.3) | 1 | 0.01 | 1 | 0.02 | |
| 100-199 | 56/67 (83.6) | 1.41 (1.10-1.80) | 1.37 (1.05-1.78) | |||
| ≥200 | 64/90 (71.1) | 1.20 (0.93-1.55) | 1.16 (0.88-1.52) | |||
| New NRTIs in switch regimen | ||||||
| ≤1 | 54/72 (75.0) | 1.06 (0.90-1.26) | ||||
| ≥2 | 98/139 (70.5) | 1 | 0.48 | |||
| Age at switch | ||||||
| <35 | 65/90 (72.22) | 1.07 (0.83-1.38) | ||||
| 35-44 | 62/84 (73.81) | 1.09 (0.84-1.41) | ||||
| ≥45 | 25/37 (67.57) | 1 | 0.80 | |||
| Gender | ||||||
| Male | 60/82 (73.2) | 1 | 0.77 | |||
| Female | 92/129 (71.3) | 0.97 (0.82-1.16) | ||||
| Reported non-adherence, second-line ART c | ||||||
| Yes | 1/3 (33.33) | 0.19 (0.02-2.12) | ||||
| No | 151/208 (72.60) | 1 | 0.16 | |||
a
Wald test; b only one patient switched regimens for non-adherence in the community programme. This patient has therefore been excluded as it would not be possible to assess a potential association); c not included in the multivariable model as considered to be on the causal pathway between exposures at time of switch and viral suppression on second-line ART.