Revising the Decision Analysis For Febrile Children At Risk For Occult Bacteremia In a Future Era of Widespread Pneumococcal Immunization Loren G. Yamamoto, MD, MPH, MBA University of Hawaii John A. Burns School of Medicine Kapiolani Medical Center For Women And Children

Febrile Children at Risk for Occult Bacteremia 3 to 36 months Temp > 39 C (102.2 F) No reliable source of fever identified Looks good clinically

Strategies (X) Obs: Observation (Y) CBC: CBC first. If WBC high, then BC+Abx. If WBC low, observe. (Z) BC+Abx: Blood culture and empiric antibiotics for all. Abx: Antibiotics without testing.

Decision Analysis Compares outcome variables of the different strategies Based on multiple assumptions What outcome variable should be used? Death and neurologic disability

Previous Decision Analyses Kramer: Observation superior Lieu: BC+Abx superior Downs: BC+Abx superior Yamamoto: Depends on assumption of negative consequences of unnecessary treatment

Negative Consequences of Trmt Allergic reactions from antibiotics. Pain from testing. Monetary cost. Amplification factor. Anxiety factor. Excessive antibiotic use - ? resistance

New Considerations Widespread HiB vaccine Widespread pneumococcal vaccine

HiB vaccine HiB essentially eliminated Only one serotype Overall reduction in morbidity and mortality Does not change the RANKING of the strategies X, Y, Z

Pneumococcal vaccine Highly efficacious in preliminary studies Covers limited number of serotypes Probably reduces risk of bacteremia Likely to become widespread in the near future

Decision analysis assumptions Multiple Beyond the scope of this presentation Standard assumptions which have been used in previous decision analysis

Overall morbidity/mortality reduced For strategies X (Obs), Y (CBC+) and Z (BC+Abx), bad outcomes per 100,000: Pre-HiB vaccine era: 76, 36, 17 Post-HiB vaccine era: 37, 25, 15 Future era of widespread pneumococcal vaccine (at 80% efficacy): 15, 10, 13

Conclusions Overall morbidity/mortality reduced RANKING of strategies largely unaffected by pneumococcal vaccine efficacy, but more so dependent on negative consequences of treatment assumption. The magnitude of the differences between strategies is smaller.

Conclusions Since the difference between strategies X, Y, Z are small, consider them similar. Social factors surrounding the patient encounter may be important in selecting a strategy

Social Factors Patient observation and follow-up reliability. Parents demand tests and antibiotics. Parents’ fear of tests. Parents are lawyers, physicians, etc. Primary care office versus emergency department.