Longitudinal effect of CD4 by cotrimoxazole use on malaria incidence among HIV-infected Ugandan adults on antiretroviral therapy: a randomized controlled study

Table 2 Effect of trial drug on malaria, by CD4 count at enrolment, ART initiation and at the time of malaria episode

Trial arm	Stratum	Episodes	Person years	Rate/100 person years (95 % CI)^a	Rate ratio^b	P value^c
CD4 count at ART initiation
CTX	<100	30	698	4.0 (2.8–5.9)	1	0.99
	100–249	59	1461	3.9 (3.0–5.1)	1.0 (0.6–1.5)
	250+	10	229	3.9 (2.0–7.4)	1.0 (0.5–2.0)
Placebo	<100	92	709	12.4 (9.8–15.7)	1	0.73
	100–249	196	1458	13.4 (11.4–15.7)	1.1 (0.8–1.4)
	250+	36	212	14.7 (10.1–21.5)	1.2 (0.8–1.9)
CD4 count at enrolment
CTX	<350	14	416	3.5 (2.0–5.9)	1	0.51
	350–499	32	900	3.4 (2.4–4.8)	1.0 (0.5–1.9)
	500+	57	1234	4.3 (3.3–5.7)	1.3 (0.7–2.3)
Placebo	<350	58	453	13.1 (9.8–17.5)	1	0.98
	350–499	117	850	13.6 (11.1–16.7)	1.0 (0.7–1.5)
	500+	175	1211	13.4 (11.3–16.0)	1.0 (0.7–1.4)
CD4 count at infection
CTX	<350	21	4618	4.4 (2.9–6.9)	1	0.52
	350–499	30	8811	3.3 (2.3–4.7)	0.7 (0.4–1.3)
	500+	52	1197	4.0 (3.0–5.4)	0.9 (0.5–1.5)
Placebo	<350	60	390	14.6 (11.1–19.2)	1	0.32
	350–499	98	807	11.8 (9.5–14.6)	0.8 (0.6–1.1)
	500+	192	1317	14.1 (12.0–16.5)	1.0 (0.7–1.3)

^aMarginal means from random effects Poisson regression model with fixed effects for CD4 count stratum, treatment arm and their interaction, and site and year since enrolment
^bRate ratio for effect of treatment arm in each CD4 count stratum, adjusted for site and year since enrolment, from random effects Poisson regression model
^cP values for overall association of CD4 count with malaria incidence within each treatment arm. P values for interaction between CD4 count and treatment arm: CD4 count at ART initiation P = 0.87; CD4 count at enrolment P = 0.60; CD4 count at infection P = 0.96

ISSN: 1475-2875