Strategies for monitoring and evaluation of resource-limited national antiretroviral therapy programs: the two-phase design

Table 4 Estimated odds ratios (OR) and 95% confidence intervals (CI) from logistic regression models for a negative outcome status at six months, based on five scenarios for patient and aggregated data availability

	Complete patient data (N = 82,887)		Aggregated quarterly-clinic data (N = 1,518)		Case–control study (N = 5,000)		Two-phase study^d
	Complete patient data (N = 82,887)		Aggregated quarterly-clinic data (N = 1,518)		Case–control study (N = 5,000)		Design #1 (N = 5,000)		Design #2 (N = 5,000)
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Age ^a
Linear	0.96	(0.94, 0.98)	0.35	(0.19, 0.66)	0.94	(0.25, 0.52)	0.94	(0.87, 1.01)	0.88	(0.80, 0.95)
Quadratic	1.06	(1.05, 1.07)	0.77	(0.54, 1.10)	1.05	(0.88, 1.00)	1.04	(0.99, 1.08)	1.03	(0.98, 1.08)
Gender
Male	REF		REF		REF		REF		REF
Female^b	0.72	(0.69, 0.74)	0.87	(0.82, 0.93)	0.75	(0.66, 0.84)	0.76	(0.67, 0.87)	0.71	(0.62, 0.82)
WHO stage
1/2	REF		REF		REF		REF		REF
3/4^c	3.33	(2.97, 3.74)	1.57	(1.44, 1.72)	3.69	(2.62, 5.20)	3.31	(2.41, 4.55)	4.31	(2.87, 6.46)
Region
Central/North	REF		REF		REF		REF		REF
South	0.92	(0.88, 0.95)	0.88	(0.82, 0.93)	0.93	(0.83, 1.05)	0.91	(0.80, 1.04)	1.02	(0.88, 1.17)
Year of registration
2005	REF		REF		REF		REF		REF
2006	0.91	(0.87, 0.95)	1.00	(0.94, 1.07)	0.97	(0.83, 1.05)	0.83	(0.68, 1.01)	0.88	(0.83, 0.93)
2007	0.76	(0.73, 0.80)	0.91	(0.85, 0.97)	0.74	(0.63, 0.86)	0.75	(0.62, 0.91)	0.75	(0.71, 0.80)
Clinic type
Public	REF		REF		REF		REF		REF
Private	0.31	(0.20, 0.48)	1.08	(0.99, 1.19)	0.39	(0.13, 1.14)	0.31	(0.19, 0.50)	0.31	(0.20, 0.47)
Clinic/year interaction
Private/2006	2.16	(1.35, 3.46)	2.03	(0.95, 4.35)	2.20	(0.67, 7.28)	2.18	(1.30, 3.67)	2.20	(1.37, 3.53)
Private/2007	2.05	(1.26, 3.32)	1.98	(0.91, 4.32)	2.36	(0.69, 8.05)	2.07	(1.21, 3.52)	1.98	(1.22, 3.21)

^aQuarterly-clinic cohort data model uses average age (in years).
^bOR for the quarterly-clinic cohort data model corresponds to a contrast of 20% in the percent female.
^cCovariate in the quarterly-clinic cohort data model is a binary indicator of whether or not the the percent WHO stage 3/4 is ≤/> 90%.
^dSee Table 2 for details; Design #1 stratifies on clinic type; Design #2 stratifies on clinic type and year of registration.

ISSN: 1471-2288