The comparative cost-effectiveness of statistical decision rules and experienced physicians in pharyngitis management

Area of Science:

• Medical Decision Making
• Health Economics
• Infectious Disease Epidemiology

Background:

• Streptococcal pharyngitis diagnosis relies on clinical judgment and throat cultures.
• Physician decision-making may not always align with optimal cost-effectiveness.
• Predictive models offer a data-driven approach to clinical decisions.

Purpose of the Study:

• To compare the cost-effectiveness of probability-based decision rules versus physician decisions for streptococcal pharyngitis.
• To evaluate different predictive models in guiding diagnostic and treatment strategies.

Main Methods:

• Retrospective analysis of 310 patients with streptococcal pharyngitis.
• Utilized four predictive models (discriminant analysis, branching algorithm, logistic regression) to calculate disease probability.
• Applied Tompkins' decision rules to projected decisions based on model probabilities.
• Calculated direct medical and indirect costs per correct action (treated diseased patients, untreated non-diseased patients).

Main Results:

• Two predictive models demonstrated greater cost-effectiveness than the ten physicians' actual decisions.
• Model 1 primarily reduced treatment costs while ensuring no diseased patients were untreated.
• Model 4 significantly reduced throat culture costs but resulted in 15% projected undertreatment.

Conclusions:

• Statistical decision rules, when integrated with Tompkins' rules, show potential for cost-effective management of streptococcal pharyngitis.
• Model 1's approach balances cost reduction with effective treatment, while Model 4 prioritizes reducing diagnostic costs.
• Adoption of statistical decision rules should consider physician and patient priorities alongside economic factors.