Skip to main content

Table 3 Final multivariable model showing factors associated with viral suppression among refugees and host nationals on ART for ≥25 weeks at baseline in Kakuma, Kenya (N = 128a)

From: Low levels of viral suppression among refugees and host nationals accessing antiretroviral therapy in a Kenyan refugee camp

Factor

Prevalence <5000 copies/mL, n/N (%)

Crude odds ratio (95% CI)

p-value

Adjusted odds ratio (95% CI)b

p-value

Age group (years)c

 18+

14/31 (45)

1

p(tr) = 0.90

1

p(tr) = 0.82

 30+

33/63 (52)

1.34 (0.56, 3.17)

 

1.47 (0.57, 3.76)

 

 40+

16/34 (47)

1.08 (0.41, 2.87)

 

0.90 (0.31, 2.59)

 

Refugee status

 Host

9/20 (45)

1

p = 0.08

1

p = 0.17

 Refugee

19/41 (46)

1.88 (0.93, 3.81)

 

1.69 (0.79, 3.57)

 

Time on ART (years)c

35/67 (52)

    

 0-

 

1

p(tr) = 0.49

1

p(tr) = 0.76

 1-

30/71 (42)

1.06 (0.36, 3.09)

 

0.80 (0.26, 2.49)

 

 2+

33/57 (58)

1.34 (0.49, 3.64)

 

1.00 (0.34, 2.95)

 

Time from HIV diagnosis to ART start (weeks)c

 0-

8/29 (28)

1

p(tr) = 0.006

1

p(tr) = 0.01

 24-

6/13 (46)

2.25 (0.58, 8.78)

 

2.71 (0.66, 11.11)

 

 48+

49/86 (57)

3.48 (1.39, 8.72)

 

3.61 (1.37, 9.47)

 

Adherence to medication schedule, self-reportedb

 Inconsistent

7/25 (28)

1

p = 0.02

1

p = 0.02

 Consistent

56/103 (54)

3.06 (1.18, 7.96)

 

3.12 (1.14, 8.49)

 

Self-reported dosing scheduleb, d

 Incorrect dosing

11/28 (39)

1

p = 0.23

1

p = 0.05

 Correct dosing

52/100 (52)

1.67 (0.71, 3.93)

 

2.52 (0.96, 6.58)

 
  1. p-values are log likelihood ratio tests; CI, confidence interval
  2. aThree clients with incomplete data were excluded
  3. bAdjusted for all factors in table except adherence factors denoted by superscript d. Factors were not adjusted for adherence to avoid risk of over-adjustment bias given that adherence mediates viral load
  4. cFactor modelled as a linear effect; p(tr) = p(trend)
  5. dIncorrect dosing was determined by comparing self-reported dosing schedules to standard dosing schedules [45]