We prospectively studied the demographics, the clinical and diagnostic
features, the HIV-1 serostatus and the therapeutic response for all n
ew patients with septic arthritis (SA) admitted to the Department of I
nternal Medicine of the Centre Hospitalier de Kigali, Rwanda, over a 1
9 month period. SA was diagnosed in 24 patients (10 male, 14 female),
of whom 19 (79%) were HIV-1 seropositive (HIVpos). Gonococcal arthriti
s was found in four patients, all HIVpos. Non-gonococcal bacterial art
hritis was established in 16 patients, of whom 13 were HIVpos. Causati
ve organisms involved in this group and the corresponding HIV-1 serost
atus of the patients were: Staphylococcus aureus: 4; 2 HIVpos, 2 HIVne
g; Streptococcus pneumoniae: 4; 4 HIVpos; Salmonella group B: 2; 2 HIV
pos; Streptococcus group D: 1; 1 HIVpos; Klebsiella pneumoniae: 1; 1 H
IVpos; undetermined: 4; 3 HIVpos, 1 HIVneg. Tuberculous arthritis was
presumed in four patients, of whom two were HIVpos. HIV-l-associated S
A had a classical acute presentation and an overall good prognosis. Co
mpared to a control group consisting of hospitalized patients with mal
aria as the sole diagnosis, patients with SA were more likely to be in
fected with HIV-1 (P = 0.005, OR 6.3; 95% CI 1.7-22.2). Prevalence rat
e estimates of SA among HIVpos and HIVneg patients were 0.5 and 0.25%,
respectively (P = 0.38). We conclude that HIV-1 infection appears as
a risk factor for SA among patients hospitalized at the Centre Hospita
lier de Kigali, but that SA cannot be used a a predictor for HIV-1 inf
ection for hospitalized patients. SA occurs infrequently and may prese
nt at any stage of HIV-1 infection.