Background:  ART programs in Sub-Saharan Africa report a high mortality in the first year of ART. There remains debate as to the public health importance of immune reconstitution inflammatory syndrome (IRIS), particularly whether IRIS is a significant contributor to this early mortality. Our objective was to determine the contribution of IRIS to mortality, morbidity, hospitalization, and treatment changes in the first 6 months of ART at 2 sites in South Africa.

Methods:  Prospective cohort study of patients initiating ART between December 2006 and October 2007 at 2 hospital-based clinics in Durban, South Africa. Patients were monitored monthly for 6 months for clinical events using structured symptom questionnaires. All new and worsening events were assessed by ≥2 expert physicians and classified as IRIS, possible IRIS or non-IRIS.

Results:  A total of 498 patients initiated ART (stavudine [d4T], lamivudine [3TC], and nevirapine [NVP; 37%] or efavirenz [EFV; 63%]):  25% were male, median age 35 years, median CD4⁺ count 106 cells/mL, 56% WHO stage 3 and 13% stage 4. IRIS accounted for 136 of 632 (22%) of all clinical events, and a further 137 of 632 (22%) events were classified as possible IRIS. Cumulative incidence of IRIS at 6 months was 22.5% (rate 58.9 /100 person-years), and all-cause mortality was 5.6%. Most IRIS events were dermatological (66%), including itchy folliculitis/papular pruritic eruption (29%), genital ulcers (10%), warts (8%), and shingles (7%). More serious IRIS events were due to TB (24%) or cryptococcosis (2%), with one case each of Kaposi’s sarcina, strongyloidiasis, sarcoidosis, and non-TB mycobacterial infection. Of serious events, 58% were unmasking (new presentation of disease). Of 28 deaths, 5 (17.9%) were due to IRIS (TB [2], cryptococcosis [1], Kaposi’s sarcoma [1], and strongyloidiasis [1]); all occurred in patients with baseline CD4 <60 cells/mL. IRIS resulted in 14 of 70 (20%) hospital admissions and 10 of 36 (28%) ART changes or interruptions (all switches from NVP to EFV because of rifampicin interaction). Non-fatal hospitalizations related to IRIS were due to TB (8), cryptococcosis (1), and severe genital warts (1).

Conclusions:  IRIS accounted for nearly a fifth of all deaths, 20% of hospital admissions, and 28% of treatment changes in an out-patient ART program in the first 6 months of ART. The impact of IRIS is likely to increase as ART programs in Africa are integrated with in-patient services, and ART is increasingly offered to those with opportunistic infections and more advanced disease.