Table 3.
Univariable analysis for association with percentage of ART and PMTCT coverage (logit-transformed; based on UNAIDS estimates)
| ART coverage (n=678) | PMTCT coverage (n=348) | |||
|---|---|---|---|---|
|
|
|
|||
| Adjusted OR (95% CI) | P (Wald χ2)e | Adjusted OR (95% CI) | P (Wald χ2)e | |
| Base model | ||||
| Year | 1.65 (1.55, 1.76) | <0.001 (248) | 1.49 (1.37, 1.61) | <0.001 (90.4) |
| HIV prevalence, % | ||||
| Linear | 0.86 (0.73, 1.00)b | 0.05 (3.7)b | 0.88 (0.70, 1.12)b | 0.31 (1.1)b |
| Quadratic | 1.01 (1.00, 1.01) b | 0.04 (4.1) b | 1.01 (1.00, 1.02)b | 0.17 (1.9)b |
| Random effects | ||||
| Intercept (or country) | – | <0.001 (524) c | – | <0.001 (268) c |
| Year | – | <0.001 (159) d | – | 0.02 (8.21) d |
| Health variables | ||||
| log(TB DALY/10,000 in 2004) | 0.35 (0.24, 0.50) | <0.001 (31.4) | 0.26 (0.14, 0.48) | <0.001 (18.5) |
| log(maternal and child DALY/10,000 in 2004) | 0.38 (0.24, 0.62) b | <0.001 (15.4) b | 0.13 (0.06, 0.28) | <0.001 (27.4) |
| log(non-HIV DALY/10,000 in 2004) | 0.09 (0.03, 0.27) | <0.001 (18.5) | 0.02 (<0.01, 0.14) | <0.001 (16.6) |
| log(international non-HIV health funding, USC|/capita) | 1.68 (1.35, 2.10) | <0.001 (20.9) | 1.06 (0.67, 1.69) | 0.80 (0.07) |
| log(international HIV funding, USC|/ HIV-infected person) | 0.97 (0.85, 1.12)b | 0.71 (0.13)b | 1.00 (0.77, 1.30)b | 0.99 (<0.01)b |
| Received Global Fund for HIV | 1.02 (0.84, 1.23) | 0.87 (0.03) | 1.00 (0.68, 1.46) | 0.99 (<0.01) |
| Received PEPFAR focus country funding | 1.52 (1.06, 2.18) | 0.02 (5.1) | 0.86 (0.47, 1.58) | 0.63 (0.24) |
| Skilled birth attendants at delivery (per 10%) | 1.11 (1.01, 1.21) | 0.03 (4.9) | 1.30 (1.12, 1.52) | 0.001 (11.2) |
| Pregnant women receiving antenatal care (per 10%) | 1.15 (1.02, 1.30) | 0.02 (5.2) | 1.27 (0.97, 1.65) | 0.08 (3.06) |
| Socio-economic and geographical variables | ||||
| Log(GDP/capita) | 3.38 (1.92, 5.95) b | <0.001 (17.9) b | 11.6 (4.78, 28.2) b | <0.001 (29.4) b |
| Adult literacy rate (per 10%) | 1.19 (1.06, 1.34) b | 0.003 (8.5) b | 1.40 (1.15, 1.72) b | <0.001 (10.6) b |
| Access to sanitation (per 10%) | 1.11 (1.02, 1.21) b | 0.02 (5.4) b | 1.29 (1.12, 1.49) b | <0.001 (12.8) b |
| Gender inequality indexa | 0.80 (0.62, 1.03)b | 0.08 (3.2)b | 0.61 (0.41, 0.90) b | 0.01 (6.3) b |
| Ethno-linguistic fractionalizationa | 0.85 (0.78, 0.92) | <0.001 (17.5) | 0.87 (0.76, 1.00) | 0.04 (4.1) |
| Regions (reference: Asia-Pacific) | <0.001 (28.2) | <0.001 (33.6) | ||
| Europe and Central Asia | 1.05 (0.50, 2.22) | 0.89 (0.02) | 3.45 (0.97, 12.2) | 0.06 (3.7) |
| Latin America and the Caribbean | 3.71 (1.90, 7.24) | <0.001 (14.8) | 9.53 (3.04, 29.9) | <0.001 (14.9) |
| Middle-East and North Africa | 0.68 (0.26, 1.79) | 0.44 (0.61) | 0.18 (0.03, 0.99) | 0.05 (3.9) |
| Sub-Saharan Africa | 0.85 (0.41, 1.76) | 0.65 (0.20) | 0.97 (0.30, 3.18) | 0.96 (<0.01) |
| Political variables | ||||
| Democracy score | 1.10 (1.04, 1.17) b | 0.001 (10.3) b | 1.23 (1.09, 1.38) | <0.001 (12.5) |
| Political voice and accountability | 2.05 (1.62, 2.60) | <0.001 (35.1) | 2.06 (1.35, 3.14) | <0.001 (11.1) |
| Political stability | 1.96 (1.62, 2.36) | <0.001 (48.1) | 1.72 (1.23, 2.41) | <0.001 (10.0) |
| Political control of corruption | 2.64 (1.95, 3.55) b | <0.001 (40.3) b | 3.14 (1.89, 5.22) | <0.001 (19.5) |
| Rule of law | 1.81 (1.35, 2.45) | <0.001 (15.3) | 2.68 (1.61, 4.46) | <0.001 (14.3) |
| Government effectiveness | 1.83 (1.35, 2.48) | <0.001 (15.1) | 3.33 (2.04, 5.43) | <0.001 (23.2) |
| Regulatory quality of government | 1.37 (1.04, 1.82) b | 0.03 (4.9) b | 2.27 (1.42, 3.63) | 0.001 (11.7) |
a
Rescaled so that the plausible range is from 0 to 10 (original plausible range was from 0 to 1).
b
Missing data are imputed for these variables; χ2 statistic is estimated from (F statistic)×(dfnum), where the numerator degrees of freedom, dfnum=1 for continuous covariates, is given instead.
c
Likelihood ratio test of multilevel model with random intercept only against linear regression with fixed effects only.
d
Likelihood ratio test of multilevel model with random intercept and slope of year against multilevel model with random intercept only.
e
Values in bold indicate statistical significance at p<0.05.