Objective: To study the incidence, etiology, and outcome of pulmonary infiltrates (PIs) in HIV-infected patients and to evaluate the yield of diagnostic procedures.
Design: Prospective observational study of consecutive hospital admissions.
Setting: Tertiary hospital.
Patients: HIV-infected patients with new-onset radiologic PIs from April 1998 to March 1999.
Methods: The study protocol included chest radiography, blood and sputum cultures, serologic testing for "atypical" causes of pneumonia, testing for Legionella urinary antigen, testing for cytomegalovirus antigenemia, and bronchoscopy in case of diffuse or progressive PIs.
Results: One hundred two episodes in 92 patients were recorded. The incidence of PIs was 18 episodes per 100 hospital admission-years (95% confidence interval [CI]: 15-21). An etiologic diagnosis was achieved in 62 cases (61%). Bacterial pneumonia (BP), Pneumocystis carinii pneumonia (PCP), and mycobacteriosis were the main diagnoses. The incidences of BP and mycobacteriosis were not statistically different in highly active antiretroviral therapy (HAART) versus non-HAART patients. The incidence of PCP was lower in those receiving HAART (p =.011), however. Nine patients died (10%). Independent factors associated with higher mortality were mechanical ventilation (odds ratio [OR] = 83; CI: 4.2-1,682), age >50 years (OR = 23; CI: 2-283), and not having an etiologic diagnosis (OR = 22; CI: 1.6-293).
Conclusions: Pulmonary infiltrates are still a frequent cause of hospital admission in the HAART era, and BP is the main etiology. There was no difference in the rate of BP and mycobacteriosis in HAART and non-HAART patients. Not having an etiologic diagnosis is an independent factor associated with mortality.