Suspected Brazilian Purpuric Fever, Brazilian Amazon Region

To the Editor: Brazilian purpuric fever (BPF), a Haemophilus aegyptius–caused febrile hemorrhagic illness of children that begins with conjunctivitis and has a case-fatality rate of 40%–90% (1,2), was fi rst recognized during a 1984 outbreak. Before June 2007, 69 cases were reported worldwide; 65 were from Brazil (1–3). To our knowledge, the disease had not been reported in the Amazon region until this investigation, which was precipitated by the report of 5 cases of a compatible syndrome in Anajás, Pará State, Brazil, in August 2007. To determine whether recent reports of BPF were accurate, we reviewed medical records of the hospital in Anajás. We identifi ed cases by using the following defi nition: fever >38.5oC , abdominal pain, vomiting, purpura, and antecedent conjunctivitis during July 1–September 30, 2007, in a child 3 months–10 years of age; absence of signs or symptoms of meningitis in those children; and laboratory exclusion of meningococcal infection. In addition, we searched retrospectively and prospectively for conjunctivitis among pupils of the elementary schools of Anajás during July–September 2007. We found 7 children with illnesses that met our case defi nition. From 2 children with nonfatal illness, we collected blood, serum, conjunctival swabs, and cerebrospinal fl uid (CSF). All specimens were submitted for bacterial culture in half agar chocolate without bacitracin; serum and CSF were also subjected to real-time PCR for detection of Neisseria meningitides, Streptococcus pneumoniae, and Haemophilus infl uenzae serotypes a, b, c, and d and to conventional PCR for the ompP2 gene of H. infl uenzae. All serum samples were also tested by hemagglutination inhibition for Flavivirus, Oropouche, Catu, Caraparu, Tacaiuma, Mayaro, Mucambo, western equine encephalitis, eastern equine encephalitis, Guaroa, Maguari, Ilhéus, Rocio, and St. Louis encephalitis; by immunoglobulin M antibody-capture ELISA for dengue and yellow fever; and, when reactive for dengue, by reverse transcription– PCR for dengue types 1, 2, 3, and 4. Because of the remoteness of the outbreak site, samples for bacterial culture were collected on locally available blood agar enriched with rabbit serum without antimicrobial drug–selective agents, rather than on the recommended chocolate agar enriched with horse serum and bacitracin (1). Samples were transported over several days by open boat at ambient temperature (≈35 oC) in improvised containers without an incubator. Serum and CSF samples were thawed and refrozen repeatedly for removal of aliquots before testing. Microbiologic and virologic testing was conducted at the Pará State Health Laboratory and Evandro Chagas Institute. Serum and CSF samples were tested by PCR at Adolfo Lutz Institute. We identifi ed 7 case-patients (median age 4 years, range 2–8 years): 6 from review of charts at the local hospital and 1 from active search in the rural community. Onset of illness was August 1 for the fi rst case-patient and August 31 for the last. Five (71%) did not receive antimicrobial drugs and died within 24 hours after fever onset; 2 were treated with amoxicillin within 24 hours after fever onset and survived (Table). Laboratory tests showed leukopenia on the day of hospital admission (Table). All case-patients had antecedent conjunctivitis. All except the fi rst case-patient had had physical contact with a previous case-patient through school or family; 5 were related (siblings or cousins). The period from exposure to onset of fever was 8–21 days. Of 1,598 elementary school pupils investigated for conjunctivitis, 111 (7%) reported symptoms of conjunc-