Diagnostic testing and discharge coding for whooping cough in a children's hospital

To evaluate the diagnostic pathways for whooping cough in a large urban paediatric hospital to inform assessment of the relative merits of notification and hospitalization data for measuring pertussis disease burden in Australian children.

All laboratory requests for Bordetella pertussis (BP) culture or serology between 30 June 1997 and 30 June 1999 were reviewed and cross-checked against discharge diagnoses with International Classification of Disease (ICD) codes A37.0, 033.0 (whooping cough due to BP) or 37.9, 033.9 (whooping cough due to unspecified organisms). Culture-positive (CP) cases were defined as a positive culture or polymerase chain reaction for BP. Culture-negative (CN) cases either fulfilled the current Australian clinical case definition (>/=14 days of cough with one or more of paroxysms, whoop, post-tussive vomiting), or had a cough illness with either positive BP serology or documented contact with an individual coughing for >14 days. In infants <6-months-old, a coughing illness with apnoea and negative investigations for other causes was also accepted. Culture positive and CN cases were cross-referenced with notification data.

During the study period, laboratory tests for BP were performed in 677 children, of whom 230 were hospitalized and 71 (31%) had an eligible ICD code at discharge; 29 were CP, 40 CN, and two (3%) were misclassified. A further 14 CP children were not admitted. Although 61 hospitalized cases (88%) fulfilled notification criteria, including 32 (80%) of CN cases, only 26 (90%) of CP and eight (20%) of CN cases were notified.

Notifications substantially under-enumerate hospitalized infant cases, especially those without positive laboratory tests. Hospital discharge data add significantly to surveillance for pertussis, particularly in infancy where most severe cases occur.