Population-level adult mortality following the expansion of antiretroviral therapy in Rakai, Uganda

Figures & data

Table 1 Person-years, deaths, and mortality rates by sex and treatment period, ages 15–49, Uganda 2000–14

Variables	Pre-ART period (2000–03)^1		ART period (2004–14)^2
	Male	Female	Male	Female
Number of individuals	9,396	10,776	17,999	20,700
Number of person-years	20,924	22,822	67,403	72,432
Number of deaths	249	313	432	380
Number of deaths to HIV-positive individuals	125	224	185	193
Mortality rates (per 1,000 pyo)
Overall death rate (all-cause)	11.9 (10.5–13.5)	13.7 (12.3–15.3)	6.4 (5.8–7.0)	5.2 (4.7–5.8)
Death rate for HIV-negative individuals	3.1 (2.2–4.3)	2.3 (1.6–3.3)	3.0 (2.4–3.6)	2.2 (1.8–2.7)
Death rate for HIV-positive individuals (PLHIV)	101.0 (84.7–120.3)	100.4 (88.1–114.4)	43.9 (38.0–50.7)	24.1 (20.9–27.7)
Death rate for individuals with HIV status unknown	10.6 (8.6–13.0)	8.5 (6.6–11.1)	5.1 (4.3–6.0)	4.3 (3.5–5.2)

¹The period 2000–03 includes January–May 2004.

²The period 2004–14 excludes January–May 2004.

Note: 95 per cent confidence intervals are shown in parentheses. Pyo refers to person-years of observation.

Source: Authors’ calculations from RCCS data.

Figure 1 Life-years lived by (a) HIV-negative individuals, (b) HIV-positive individuals, and (c) individuals with unknown HIV status, compared with the population as a whole; and (d) population-wide life-years lived and lost to HIV; all between ages 15 and 50, by sex and calendar year, Uganda 2000–14

Source: Authors’ calculations from RCCS data.

Figure 2 Trends in the total number of deaths among PLHIV (left-hand axis) and all-cause mortality rates for PLHIV (right-hand axis), by treatment status, Uganda 2000–14

Source: As for Figure 1.

Table 2 All-cause mortality in HIV-positive individuals: hazard ratios and 95 per cent confidence intervals (Weibull regression), Uganda 2000–14

Variables^1	Pre-ART (2000–03)^2	ART (2004–14)^3
		Untreated	Treated
Calendar year (2000–03)	0.78	–	–
	(0.69–0.88)	–	–
Calendar year (2004–08)	–	0.72	0.64
	–	(0.67–0.77)	(0.57–0.71)
Calendar year (2009–14)	–	0.87	0.96
	–	(0.77–0.99)	(0.79–1.16)
Male^4	0.65	0.93	0.93
	(0.45–0.94)	(0.76–1.14)	(0.75–1.16)
Male × calendar year (2000–03)	1.29	–	–
	(1.05–1.57)	–	–
Male × calendar year (2004–08)	–	1.15	1.36
	–	(1.04–1.28)	(1.17–1.58)
Male × calendar year (2009–14)	–	1.08	0.76
	–	(0.92–1.28)	(0.57–1.01)
Weibull shape parameter (p)	1.90	1.79	1.75
	(1.47–2.45)	(1.47–2.18)	(1.36–2.26)
Subjects	1,914	4,465	2,534
Deaths	349	645	431

¹Estimates presented are hazard ratios from a Weibull regression model (with 95 per cent confidence intervals in parentheses). ‘Calendar year’ refers to a per-year relative hazard in the period specified.

²In the pre-ART analysis, the year 2000 is the baseline calendar year.

³In the ART analysis, we use 2000–03 as the baseline calendar year (including all individuals/episodes regardless of ultimate treatment status) for both untreated and treated analyses, and distinguish between untreated or treated episodes only from 2004–14. We do this to model the trend in mortality rates by treatment status from when ART was first introduced in Rakai.

⁴Reference group is females.

Source: As for Table 1.

Figure A1 Model-predicted hazard of mortality for HIV-positive individuals by sex, calendar year, and ART status, Uganda 2000–14

Note: Dashed vertical line at 2008 represents the spline inflection in the model. Source: As for Figure 1.